2024/03/29 更新

写真a

カマタ セイイチロウ
鎌田 清一郎
所属
理工学術院 大学院情報生産システム研究科
職名
教授
学位
博士(工学) ( 九州工業大学 )

所属学協会

  •  
     
     

    Institute of Electrical and Electronics Engineers

  •  
     
     

    映像情報メディア学会

  •  
     
     

    情報処理学会

  •  
     
     

    電子情報通信学会

研究分野

  • 通信工学 / 知能ロボティクス

研究キーワード

  • 画像情報処理、パターン認識、マルチメディア、信号処理

受賞

  • Best Paper Award

    2020年03月   ICIPRoB2020   Constant-Time Gaussian Filtering for Acceleration of Structure Similarity  

    受賞者: Tomohiro SASAKI, Norishige FUKUSHIMA, Yoshihiro MAEDA, Kenjiro SUGIMOTO, Seiichiro KAMATA

  • Best Paper Award 2019

    2019年09月   ICSIPA2019   Edge-guided Hierarchically Nested Network for Real-time Semantic Segmentation  

    受賞者: Yuqi LI, Sei-ichiro KAMATA, Haoran LIU

  • 最優秀論文賞(ICIEV2018)

    2018年06月   ICIEV2018  

  • 最優秀論文賞(映像メディア処理シンポジウム,IMPS)

    2017年09月   映像メディア処理シンポジウム  

    受賞者: 杉本憲治郎, 鎌田清一郎

  • 最優秀論文賞(ICARCV2010)

    2010年12月   ICARCV2010  

 

論文

  • Learned Image Compression With Multi-Scan Based Channel Fusion,

    Yuan Li, Weilian Zhou, Pengfeng Lu, Sei-ichiro Kamata

    30th IEEE International Conference on Image Processing    2023年10月  [査読有り]

  • AD/MCI Classification Using DMN Connectivity Networks-Based HGCN With Attention

    Zhihao Zuo, Sei-ichiro Kamata

    6th International Conference on Pattern Recognition and Artificial Intelligence    2023年08月  [査読有り]

  • Mask Refined Deep Fusion Network With Dynamic Memory for Robust RGBT Tracking

    Ce Bian, Sei-ichiro Kamata

    6th International Conference on Pattern Recognition and Artificial Intelligence    2023年08月  [査読有り]

  • Visual Question Answering Based on MRF Hypergraph Transformer

    Jiawei Lin, Sei-ichiro Kamata

    6th International Conference on Pattern Recognition and Artificial Intelligence    2023年08月  [査読有り]

  • Multi-Configuration Analysis of DenseNet Architecture for Whole Slide Image Scoring of ER-IHC

    Wan Siti Halimatul Munirah Wan Ahmad, Mohammad Faizal Ahmad Fauzi, Md Jahid Hasan, Zaka Ur Rehman, Jenny Tung Hiong Lee, See Yee Khor, Lai-Meng Looi, Fazly Salleh Abas, Afzan Adam, Elaine Wan Ling Chan, Sei-Ichiro Kamata

    IEEE Access   11   79911 - 79928  2023年07月  [査読有り]

    DOI

  • Skin Lesion Classification Based on Involution Neural Networks With Triplet++ Attention Generator

    Wenhang Ou, Sei-ichiro Kamata

    8th International Conference on Biomedical Signal and Image Processing (ICBIP2023)    2023年07月  [査読有り]

  • ADHD Classification Based on fMRI Spatial-Temporal Features Using Monofractal and Multifractal

    Mengyunqiu Zhang, Sei-ichiro Kamata

    8th International Conference on Biomedical Signal and Image Processing (ICBIP2023)    2023年07月  [査読有り]

  • Multiscanning-Based RNN–Transformer for Hyperspectral Image Classification

    Weilian Zhou, Sei-Ichiro Kamata, Haipeng Wang, Xi Xue

    IEEE Transactions on Geoscience and Remote Sensing   61   1 - 19  2023年05月  [査読有り]

    DOI

  • Recurrent Frequency-Aware Transformer for Endoscopic Video Super-Resolution

    Fuzhi Zhang, Sei-ichiro Kamata

    2023 International Conference on Image Processing and Machine Intelligence    2023年02月  [査読有り]

  • Kuzushiji Recognition System Using Triplet Feature Mixer,

    Pengfeng Lu, Sei-ichiro Kamata, Weilian Zhou

    2023 International Conference on Image Processing and Machine Intelligence    2023年02月  [査読有り]

  • Skeleton-Based Action Recognition Using Spatial-Temporal Hypergraph Networks

    Qiang Yu, Sei-ichiro Kamata

    2023 International Conference on Image Processing and Machine Intelligence    2023年02月  [査読有り]

  • Deep Residual Networks With Common Linear Multi-Step and Advanced Numerical Schemes,

    Zhengbo Luo, Weilian Zhou, Sei-ichiro Kamata, Xuehui Hu

    IEEE International Conference on Image Processing    2022年10月  [査読有り]

  • Rethinking Unified Spectral-Spatial-Based Hyperspectral Image Classification Under 3D Configuration of Vision Trasformer

    Weilian Zhou, Sei-ichiro Kamata, Zhengbo Luo, Xi Xue

    IEEE International Conference on Image Processing    2022年10月  [査読有り]

  • Multiple Mask Enhanced Transformer for Robust Visual Tracking,

    Ziyu Wang, Sei-ichiro Kamata

    2022 4th International Conference on Robotics and Computer Vision (ICRCV 2022)    2022年09月  [査読有り]

  • Extended Res-UNet With Hierarchical Inner-Modules for Liver Tumor Segmentation from CT Volumes

    Jiayin Shi, Sei-ichiro Kamata

    2022 4th International Conference on Robotics and Computer Vision (ICRCV 2022)    2022年09月  [査読有り]

  • MRI Super-Resolution using Implicit Neural Representation with Frequency Domain Enhancement

    Shuangming Mao, Seiichiro Kamata

    2022 7th International Conference on Biomedical Signal and Image Processing (ICBIP)    2022年08月  [査読有り]

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • ADHD Classification With Low-Frequency Fluctuation Feature Map Based on 3D CBAMe

    Lihua Su, Sei-ichiro Kamata

    2022 7th International Conference on Biomedical Signal and Image Processing (ICBIP)    2022年08月  [査読有り]

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Hierarchical Unified Spectral-Spatial Aggregated Transformer for Hyperspectral Image Classification

    Weilian Zhou, Sei-ichiro Kamata, Zhengbo Luo, Xiaoyue Chen

    26th International Conference on Pattern Recognition    2022年08月  [査読有り]

  • Near-Infrared Image Colorization with Weighted UNet++ and Auxiliary Color Enhancement GAN

    Sicong Zhou, Sei-ichiro Kamata

    2022 7th International Conference on Image, Vision and Computing    2022年07月  [査読有り]

  • Constructing infinite deep neural networks with flexible expressiveness while training

    Zhengbo Luo, Zitang Sun, Weilian Zhou, Zizhang Wu, Sei-ichiro Kamata

    Neurocomputing   487   257 - 268  2022年05月  [査読有り]

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • Contextual Mixing Feature Unet for Multi-Organ Nuclei Segmentation

    Xi Xue, Sei-Ichiro Kamata

    Frontiers in Signal Processing   2   1 - 11  2022年03月  [査読有り]

     概要を見る

    Nuclei segmentation is fundamental and crucial for analyzing histopathological images. Generally, a pathological image contains tens of thousands of nuclei, and there exists clustered nuclei, so it is difficult to separate each nucleus accurately. Challenges against blur boundaries, inconsistent staining, and overlapping regions have adverse effects on segmentation performance. Besides, nuclei from various organs appear quite different in shape and size, which may lead to the problems of over-segmentation and under-segmentation. In order to capture each nucleus on different organs precisely, characteristics about both nuclei and boundaries are of equal importance. Thus, in this article, we propose a contextual mixing feature Unet (CMF-Unet), which utilizes two parallel branches, nuclei segmentation branch and boundary extraction branch, and mixes complementary feature maps from two branches to obtain rich and integrated contextual features. To ensure good segmentation performance, a multiscale kernel weighted module (MKWM) and a dense mixing feature module (DMFM) are designed. MKWM, used in both nuclei segmentation branch and boundary extraction branch, contains a multiscale kernel block to fully exploit characteristics of images and a weight block to assign more weights on important areas, so that the network can extract discriminative information efficiently. To fuse more beneficial information and get integrated feature maps, the DMFM mixes the feature maps produced by the MKWM from two branches to gather both nuclei information and boundary information and links the feature maps in a densely connected way. Because the feature maps produced by the MKWM and DMFM are both sent into the decoder part, segmentation performance can be enhanced effectively. We test the proposed method on the multi-organ nuclei segmentation (MoNuSeg) dataset. Experiments show that the proposed method not only performs well on nuclei segmentation but also has good generalization ability on different organs.

    DOI

  • A Two-stage Refinement Network for Nuclei Segmentation in Histopathology Images

    Peiyi Jian, Sei-ichiro Kamata

    Proc. of 2022 4th International Conference on Image, Video and Signal Processing (IVSP2022)    2022年03月  [査読有り]

  • Rethinking ResNets: improved stacking strategies with high-order schemes for image classification

    Zhengbo Luo, Zitang Sun, Weilian Zhou, Zizhang Wu, Sei-ichiro Kamata

    Complex & Intelligent Systems   8 ( 4 ) 3395 - 3407  2022年02月  [査読有り]

     概要を見る

    Abstract

    Various deep neural network architectures (DNNs) maintain massive vital records in computer vision. While drawing attention worldwide, the design of the overall structure lacks general guidance. Based on the relationship between DNN design and numerical differential equations, we performed a fair comparison of the residual design with higher order perspectives. We show that the widely used DNN design strategy, constantly stacking a small design (usually, 2–3 layers), could be easily improved, supported by solid theoretical knowledge and with no extra parameters needed. We reorganise the residual design in higher order ways, which is inspired by the observation that many effective networks can be interpreted as different numerical discretisations of differential equations. The design of ResNet follows a relatively simple scheme, which is Euler forward; however, the situation becomes complicated rapidly while stacking. We suppose that stacked ResNet is somehow equalled to a higher order scheme; then, the current method of forwarding propagation might be relatively weak compared with a typical high-order method such as Runge–Kutta. We propose HO-ResNet to verify the hypothesis on widely used CV benchmarks with sufficient experiments. Stable and noticeable increases in performance are observed, and convergence and robustness are also improved. Our stacking strategy improved ResNet-30 by 2.15% and ResNet-58 by 2.35% on CIFAR-10, with the same settings and parameters. The proposed strategy is fundamental and theoretical and can, therefore, be applied to any network as a general guideline.

    Graphical abstract

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Classification of COVID-19 on Chest CT Scans with Higher Order Residual Network

    Hao Huang, Sei-ichiro Kamata

    2022 4th International Conference on BioMedical Technology (ICBMT 2022)    2022年02月  [査読有り]

  • Multiscanning Strategy-Based Recurrent Neural Network for Hyperspectral Image Classification

    Weilian Zhou, Sei-ichiro Kamata, Zhengbo Luo, Haipeng Wang

    IEEE Transactions on Geoscience and Remote Sensing   60   1 - 18  2022年  [査読有り]

    DOI

  • Adjoint Bilateral Filter and Its Application to Optimization-based Image Processing

    Keiichiro Shirai, Kenjiro Sugimoto, Sei-ichiro Kamata

    APSIPA Transactions on Signal and Information Processing   11 ( 1 )  2022年  [査読有り]

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • Hyperspectral Image Classification Based on Multi-stage Vision Transformer with Stacked Samples

    Xiaoyue Chen, Sei-ichiro Kamata, Weilian Zhou

    Proceedings of IEEE TENCON2021    2021年12月  [査読有り]

  • Phase-based accelerated motion magnification using image pyramid

    Tomohito Mizokami, Kenjiro Sugimoto, Sei-ichiro Kamata

    Proceedings of IEEE TENCON2021    2021年12月  [査読有り]

  • Derivative Feature and Residual Spatial Attention for Low-Light Image Enhancement

    Qihan Li, Sei-ichiro Kamata

    Proceedings of International Conference on Signal Processing Systems    2021年11月  [査読有り]

  • Unsupervised Learning for Stereo Depth Estimation using Efficient Correspondence Matching

    Wenbin Hui, Sei-ichiro Kamata

    Proceedings of International Conference on Advances on Image Processing    2021年11月  [査読有り]

  • A Fast and Accurate Point Pattern Matching Algorithm based on Multi-Hilbert Scans

    Jegoon Ryu, Sei-ichiro Kamata

    Proceedings of Asian Conference on Pattern Recognition (ACPR2021)    2021年11月  [査読有り]

  • Generic Sparse Graph Based Convolutional Networks for Face Recognition

    Renjie Wu, Sei-ichiro Kamata

    Proceedings of International Conference on Image Processing    2021年09月  [査読有り]

  • Deep Neural Networks with Flexible Complexity While Training Based on Neural Ordinary Differential Equations

    Zhengbo Luo, Sei-ichiro Kamata, Zitang Sun, Weilian Zhou

    ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)     1690 - 1694  2021年06月  [査読有り]

    DOI

  • Sub-Band Grouping Spectral Feature-Attention Block for Hyperspectral Image Classification

    Weilian Zhou, Sei-ichiro Kamata, Zhengbo Luo

    ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)     1820 - 1824  2021年06月  [査読有り]

    DOI

  • Data Augmentation for Ancient Characters via Blend-Font Net

    Xiaolu Ren, Sei-Ichiro Kamata

    The 13th International Conference on Digital Image Processing (ICDIP 2021)   13  2021年05月  [査読有り]

  • Spatial information using CRF for brain tumor segmentation

    Yawen Chen, Sei-Ichiro Kamata, Rong Fan

    The 13th International Conference on Digital Image Processing (ICDIP 2021)   13  2021年05月  [査読有り]

  • Pulmonary Nodule Detection Using Improved Faster R-CNN and 3D Resnet

    Rong Fan, Sei-Ichiro Kamata, Yawen Chen

    The 13th International Conference on Digital Image Processing (ICDIP 2021)   13  2021年05月  [査読有り]

  • An efficient computational algorithm for Hausdorff distance based on points-ruling-out and systematic random sampling

    Jegoon Ryu, Sei-ichiro Kamata

    Pattern Recognition   114  2021年01月  [査読有り]

    担当区分:最終著者

    DOI

    Scopus

    9
    被引用数
    (Scopus)
  • Skin Lesion Classification Using Weakly-supervised Fine-grained Method

    Xi XUE, Sei-ichiro KAMATA, Daming LUO

    IEEE International Conference on Pattern Recognition (ICPR)    2021年01月  [査読有り]

  • Adaptive Image Compression Using GAN based Semantic-perceptual Residual Compensation

    Ruojing WANG, Zitang SUN, Sei-ichiro KAMATA, Weili CHEN

    IEEE International Conference on Pattern Recognition (ICPR)    2021年01月  [査読有り]

  • Multi-Scanning Based Recurrent Neural Network for Hyperspectral Image Classification

    Weilian ZHOU, Sei-ichiro KAMATA

    IEEE International Conference on Pattern Recognition (ICPR)    2021年01月  [査読有り]

  • Semantic Segmentation Refinement Using Entropy and Boundary-guided Monte Carlo Sampling and Directed Region Search

    Zitang SUN, Sei-ichiro KAMATA, Ruojing WANG, Weili CHEN

    IEEE International Conference on Pattern Recognition (ICPR)    2021年01月  [査読有り]

  • Image Tiling For Clustering To Improve Stability Of Constant-Time Color Bilateral Filtering

    Takahisa Miyamura, Norishige Fukushima, Muhammad Waqas, Kenjiro Sugimoto, Sei-ichiro Kamata

    2020 IEEE International Conference on Image Processing (ICIP)    2020年10月  [査読有り]

    DOI

  • Extending Compressive Bilateral Filtering For Arbitrary Range Kernel

    Yuto Sumiya, Norishige Fukushima, Kenjiro Sugimoto, Sei-ichiro Kamata

    2020 IEEE International Conference on Image Processing (ICIP)    2020年10月  [査読有り]

    DOI

  • Data Augmentation for Ancient Characters via Semi-MixFontGan

    YUAN Zhiyi, Sei-ichiro KAMATA

    Proc. of 4th IEEE International Conference on Imaging, Vision & Pattern Recognition (IVPR)    2020年08月  [査読有り]

  • A Coarse to Fine Framework for Multi-organ Segmentation in Head and Neck Images

    Yan PU, Sei-ichiro KAMATA

    Proc. of 4th IEEE International Conference on Imaging, Vision & Pattern Recognition (IVPR)    2020年08月  [査読有り]

  • Stain-Refinement and Boundary-Enhancement Weight Maps for Multi-organ Nuclei Segmentation

    Ruochan WANG, Sei-ichiro KAMATA

    Proc. of 4th IEEE International Conference on Imaging, Vision & Pattern Recognition (IVPR)    2020年08月  [査読有り]

  • Second-Order Estimation Based Attention Network for Metric Learning

    Zeyu SUN, Sei-ichiro KAMATA

       2020年08月  [査読有り]

  • Combined Convolutional Neural Network for Highly Compressed Images Denoising

    Binying LIU, Sei-ichiro KAMATA

    Proc. of 4th IEEE International Conference on Imaging, Vision &Pattern Recognition (IVPR)    2020年08月  [査読有り]

  • Diabetic retinopathy grading based on Lesion correlation graph

    Daming LUO, Sei-ichiro KAMATA

    Proc. of 4th IEEE International Conference on Imaging, Vision & Pattern Recognition (IVPR)    2020年08月  [査読有り]

  • Constant-Time Gaussian Filtering for Acceleration of Structure Similarity

    Tomohiro SASAKI, Norishige FUKUSHIMA, Yoshihiro MAEDA, Kenjiro SUGIMOTO, Sei-ichiro KAMATA

    Proc. of International Conference on Image Processing and Robotics (ICIPRoB2020)    2020年03月  [査読有り]

  • Video Super-Resolution Using Wave-Shape Network

    Yanan Wu, Sei-ichiro Kamata

    Proceedings of the 3rd International Conference on Video and Image Processing    2019年12月  [査読有り]

    DOI

    Scopus

  • Hybrid Featured based Pyramid Structured CNN for Texture Classification

    Haoran Liu, Sei-Ichiro Kamata, Yuqi Li

    2019 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)    2019年09月  [査読有り]

    DOI

  • Edge-guided Hierarchically Nested Network for Real-time Semantic Segmentation

    Yuqi Li, Sei-Ichiro Kamata, Haoran Liu

    2019 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)    2019年09月  [査読有り]

    DOI

  • Data Augmentation for Historical Documents via Cascade Variational Auto-Encoder

    Guanyu Cao, Sei-Ichiro Kamata

    2019 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)    2019年09月  [査読有り]

    DOI

  • SVDとタイリング法を用いた200FPSをもつ定数時間バイラテラルフィルタ

    杉本憲治郎, 福嶋慶繁, 鎌田清一郎

    Proc.IEEE international Conference on Image Processing (ICIP2019)    2019年09月  [査読有り]

  • 3次元フラクタル次元複雑性マップによるADHDの構造MRI画像の分類

    ワン・ティアンイー, 鎌田清一郎

    Proc.IEEE international Conference on Image Processing (ICIP2019)     1 - 6  2019年09月  [査読有り]

  • 画像トラッキングのためのランキングによるアテンションアプローチ

    ペン・シェンフイ, 鎌田清一郎, ブレッコン・トビー

    Proc.IEEE international Conference on Image Processing (ICIP2019)    2019年09月  [査読有り]

  • DCT-1を使った短窓を持ったガウスフィルタの高速化

    矢野貴大, 杉本憲治郎, 黒木祥光, 鎌田清一郎

    Proc.2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)   1 ( 1 ) 129 - 132  2019年03月  [査読有り]

    DOI

    Scopus

    8
    被引用数
    (Scopus)
  • 医用カラー画像に対するPCAによるガイドバイラテラルフィルタ

    鹿毛俊喜, 杉本憲治郎, 鎌田清一郎

    Proc.9th International Conference on Biomedical Engineering and Technology (ICBET 2019)   1 ( 1 ) 1 - 6  2019年03月  [査読有り]

  • 病状重症度による糖尿病網膜症のための眼底画像識別

    阪口愛紀, 鎌田清一郎

    Proc.9th International Conference on Biomedical Engineering and Technology (ICBET 2019)   1 ( 1 ) 1 - 6  2019年03月  [査読有り]

  • Efficient Computational Scheduling of Box and Gaussian FIR Filtering for CPU Microarchitecture

    Norishige FUKUSHIMA, Yoshihiro MAEDA, Yuki KAWASAKI, Masahiro NAKAMURA, Tomoaki TSUMURA, Kenjiro SUGIMOTO, Sei-ichiro KAMATA

    Proc. Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC 2018)    2018年11月  [査読有り]

  • 3次元ボリュームデータに対するGPUフレンドリーな近似バイラテラルフィルタ

    矢野光一, 杉本憲治郎, 鎌田清一郎

    Proc.2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)   1 ( 1 ) 2054 - 2058  2018年11月  [査読有り]

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • 顔認識のためのK3-スパースグラフ・畳み込みネットワーク

    呉仁傑, 鎌田清一郎

    Proc. 2018 15th International Conference on Control, Automation, Robotics and Vision   1 ( 1 ) 174 - 179  2018年11月  [査読有り]

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Infrared Image Colorization Using a S-Shape Network

    DONG, Ziyue DONG, 鎌田 清一郎, BRECKON, Toby

    2018 25th IEEE International Conference on Image Processing (ICIP)     2242 - 2246  2018年10月  [査読有り]

    DOI

    Scopus

    30
    被引用数
    (Scopus)
  • Sparse Graph based Deep Learning Networks for Face Recognition

    WU,Renjie, 鎌田 清一郎

    IEICE Transactions on Information and Systems   E101-D ( 9 ) 2209 - 2219  2018年09月  [査読有り]

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • Nuclei Segmentation of Cervical Cell Images based on Intermediate Segment Qualifier

    WANG, Rui, 鎌田 清一郎

    IEEE Proceedings of International Conference on Pattern Recognition (ICPR2018)    2018年08月  [査読有り]

  • Deep Metric Learning with Online Hard and Soft Selection for Person Re-identification

    Yu Mingyang, 鎌田 清一郎

    Proc. Joint 7th International Conference on Informatics, Electronics & Vision (ICIEV)     426 - 431  2018年06月  [査読有り]

  • Deep Neural Networks with Mixture of Experts Layers for Complex Event Recognition from Images

    LI, Mingyao, 鎌田 清一郎

    Proc. Joint 7th International Conference on Informatics, Electronics & Vision (ICIEV)     410 - 415  2018年06月  [査読有り]

  • Character Recognition in Japanese Historical Documents via Adaptive Multi-Region Model

    WANG, Yueyu, 鎌田 清一郎

    Proc. Joint 7th International Conference on Informatics, Electronics & Vision (ICIEV)     404 - 409  2018年06月  [査読有り]

  • Frontal Gait Recognition from Incomplete RGB-D Streams Using Gait Cycle Analysis

    ZOU,Wenyun, 鎌田 清一郎

    Proc. of Joint 7th International Conference on Informatics, Electronics & Vision (ICIEV)     453 - 458  2018年06月  [査読有り]

  • Universal Approach for DCT-Based Constant-Time Gaussian Filter with Moment Preservation

    杉本 憲治郎, 京地 清介, 鎌田 清一郎

    Proc. IEEE International Conference, Acoustic, Signal Processing     1498 - 1502  2018年04月  [査読有り]

    DOI

    Scopus

    24
    被引用数
    (Scopus)
  • Guided Image Filtering with Arbitrary Window Function

    福嶋 慶繁, 杉本 憲治郎, 鎌田 清一郎

    Proc. IEEE International Conference, Acoustic, Signal Processing     1523 - 1527  2018年04月  [査読有り]

    DOI

    Scopus

    29
    被引用数
    (Scopus)
  • Copy move image forgery detection based on Polar Fourier representation

    Yitian Wang, Sei-Ichiro Kamata

    International Journal of Machine Learning and Computing   8 ( 2 ) 158 - 163  2018年04月  [査読有り]

     概要を見る

    With the rapid development of multimedia technology, it's easy for someone to obtain an image and edit it according to their own preferences or some ulterior purpose. Copy-Move is a common type of digital image forgery where a part of the original image is copied and pasted at another position in the same image. In this paper, we propose an efficient methodology for enhancing block matching based on Copy-Move forgery detection. The main contribution of this work is the utilization of polar representation to get the representative features for each block. The main feature used in this paper is the frequency of each block based on Fourier transform. The experimental results show the efficiency of the proposed method for detecting copy-move regions, even when the copied region has undergone severe image manipulations such as rotation, scaling, Gaussian blurring, brightness modification, JPEG compression and noise addition.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Complex coefficient representation for IIR bilateral filter

    Norishige Fukushima, Kenjiro Sugimoto, Sei-Ichiro Kamata

    Proceedings - International Conference on Image Processing, ICIP   2017-   2458 - 2462  2018年02月  [査読有り]

     概要を見る

    In this paper, we propose an infinite impulse response (IIR) filtering with complex coefficients for Euclid distance based filtering, e.g. bilateral filtering. Recursive filtering of edge-preserving filtering is the most efficient filtering. Recursive bilateral filtering and domain transform filtering belong to this type. These filters measure the difference between pixel intensities by geodesic distance. Also, these filters do not have separability. The aspects make the filter sensitive to noises. Bilateral filtering does not have these issues, but it is time-consuming. In this paper, edge-preserving filtering with the complex exponential function is proposed. The resulting stack of these IIR filtering is merged to approximated edge-preserving in FIR filtering, which includes bilateral filtering. For bilateral filtering, a raised-cosine function is used for efficient approximation. The experimental results show that the proposed filter, named IIR bilateral filter, approximates well and the computational cost is low.

    DOI

    Scopus

    6
    被引用数
    (Scopus)
  • Nuclei detection based on secant normal voting with skipping ranges in stained histopathological images

    Xueting Lim, Kenjiro Sugimoto, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E101D ( 2 ) 523 - 530  2018年02月

     概要を見る

    Seed detection or sometimes known as nuclei detection is a prerequisite step of nuclei segmentation which plays a critical role in quantitative cell analysis. The detection result is considered as accurate if each detected seed lies only in one nucleus and is close to the nucleus center. In previous works, voting methods are employed to detect nucleus center by extracting the nucleus saliency features. However, these methods still encounter the risk of false seeding, especially for the heterogeneous intensity images. To overcome the drawbacks of previous works, a novel detection method is proposed, which is called secant normal voting. Secant normal voting achieves good performance with the proposed skipping range. Skipping range avoids over-segmentation by preventing false seeding on the occlusion regions. Nucleus centers are obtained by mean-shift clustering from clouds of voting points. In the experiments, we show that our proposed method outperforms the comparison methods by achieving high detection accuracy without sacrificing the computational efficiency.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Supervised Two-Step Hash Learning for Efficient Image Retrieval

    Xinhui Wu, Sei-ichiro Kamata, Lizhuang Ma

    Proc. Asian Conference on Pattern Recognition     184 - 189  2017年11月  [査読有り]

  • Radical Region based CNN for Offline Handwritten Chinese Character Recognition

    Weike Luo, Sei-ichiro Kamata

    Proc. Asian Conference on Pattern Recognition     542 - 547  2017年11月  [査読有り]

  • Deep Face Recognition under Eyeglass and Scale Variation Using Extended Siamese Network

    Fan Qiu, Sei-ichiro Kamata, Lizhuang Ma

    Proc. Asian Conference on Pattern Recognition     471 - 476  2017年11月  [査読有り]

  • Face Recognition via Deep Sparse Graph Neural Networks

    Renjie WU, Sei-ichiro KAMATA, Toby BRECKON

    Proc. Workshop on Deep Learning on Irregular Domains, in British Machine Vision Conference 2017     1 - 10  2017年09月

  • Two-stage cross-based stereo disparity refinement

    Zonglin Xu, Sei-Ichiro Kamata, Qieshi Zhang

    Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017     420 - 423  2017年07月

     概要を見る

    This paper proposed a disparity refinement method based on two-stage cross. First stage is anti-texture cross-based support region construction to build proper support regions for error pixels without being influenced by texture. Based on the support regions, second stage of the method is proposed, which is called weighted cross-based updating method. The experiments show that the proposed method could build the support region accurately and improve the accuracy of the disparity map in final results with fast speed, compared to other tree-based algorithms. It also outperforms the existing disparity refinement methods in preserving the boundaries of objects in the final disparity map.

    DOI

    Scopus

  • Robust registration of serial cell microscopic images using 3D Hilbert scan search

    Yongwen Lai, Sei-Ichiro Kamata, Zhizhong Fu

    Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017     530 - 533  2017年07月

     概要を見る

    Microscopic images are quite helpful for us to observe the details of cells because of its high resolution. Furthermore it can benefit biologists and doctors to view the cell structure from any aspect by using a serial images to generate 3D cell structure. However each cell slice is placed at the microscopy respectively, which will bring in the arbitrary rotation and translation among the serial slices. What's more, the sectioning process will destroy the cell structure such as tearing or warping. Therefore we must register the serial slices before rendering the volume data in 3D. In this paper we propose a robust registration algorithm based on an improved 3D Hilbert scam search. Besides we put forward a simple but effective method to remove false matching in consecutive images. Finally we correct the local deformation based on optical-flow theory and adopt multi-resolution method. Our algorithm is tested, on a serial microscopy kidney cell images, and the experimental results show how accurate and robust of our method is.

    DOI

    Scopus

  • Discriminative Histogram Intersection Metric Learning and Its Applications

    Peng-Yi Hao, Yang Xia, Xiao-Xin Li, Sei-ichiro Kamata, Sheng-Yong Chen

    JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY   32 ( 3 ) 507 - 519  2017年05月  [査読有り]

     概要を見る

    In this paper, a novel method called discriminative histogram intersection metric learning (DHIML) is proposed for pair matching and classification. Specifically, we introduce a discrimination term for learning a metric from binary information such as same/not-same or similar/dissimilar, and then combine it with the classification error for the discrimination in classifier construction. Compared with conventional approaches, the proposed method has several advantages. 1) The histogram intersection strategy is adopted into metric learning to deal with the widely used histogram features effectively. 2) By introducing discriminative term and classification error term into metric learning, a more discriminative distance metric and a classifier can be learned together. 3) The objective function is robust to outliers and noises for both features and labels in the training. The performance of the proposed method is tested on four applications: face verification, face-track identification, face-track clustering, and image classification. Evaluations on the challenging restricted protocol of Labeled Faces in the Wild (LFW) benchmark, a dataset with more than 7 000 face-tracks, and Caltech-101 dataset validate the robustness and discriminability of the proposed metric learning, compared with the recent state-of-the-art approaches.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Eigen-aging reference coding for cross-age face verification and retrieval

    Kaihua Tang, Sei-Ichiro Kamata, Xiaonan Hou, Shouhong Ding, Lizhuang Ma

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)   10113   389 - 403  2017年

     概要を見る

    Recent works have achieved near or over human performance in traditional face recognition under PIE (pose, illumination and expression) variation. However, few works focus on the cross-age face recognition task, which means identifying the faces from same person at different ages. Taking human-aging into consideration broadens the application area of face recognition. It comes at the cost of making existing algorithms hard to maintain effectiveness. This paper presents a new reference based approach to address cross-age problem, called Eigen-Aging Reference Coding (EARC). Different from other existing reference based methods, our reference traces eigen faces instead of specific individuals. The proposed reference has smaller size and contains more useful information. To the best of our knowledge, we achieve state-of-the-art performance and speed on CACD dataset, the largest public face dataset containing significant aging information.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Constant-time bilateral filter using spectral decomposition

    Kenjiro Sugimoto, Toby Breckon, Sei-Ichiro Kamata

    Proceedings - International Conference on Image Processing, ICIP   2016-   3319 - 3323  2016年08月  [査読有り]

     概要を見る

    This paper presents an efficient constant-time bilateral filter where constant-time means that computational complexity is independent of filter window size. Many state-of-the-art constant-time methods approximate the original bilateral filter by an appropriate combination of a series of convolutions. It is important for this framework to optimize the performance tradeoff between approximate accuracy and the number of convolutions. The proposed method achieves the optimal performance tradeoff in a least-squares manner by using spectral decomposition under the assumption that images consist of discrete intensities such as 8-bit images. This approach is essentially applicable to arbitrary range kernel. Experiments show that the proposed method outperforms state-of-the-art methods in terms of both computational complexity and approximate accuracy.

    DOI

    Scopus

    14
    被引用数
    (Scopus)
  • Integrating Multiple Global and Local Features by Product Sparse Coding for Image Retrieval

    Li Tian, Qi Jia, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E99D ( 3 ) 731 - 738  2016年03月  [査読有り]

     概要を見る

    In this study, we propose a simple, yet general and powerful framework of integrating multiple global and local features by Product Sparse Coding (PSC) for image retrieval. In our framework, multiple global and local features are extracted from images and then are transformed to Trimmed-Root (TR)-features. After that, the features are encoded into compact codes by PSC. Finally, a two-stage ranking strategy is proposed for indexing in retrieval. We make three major contributions in this study. First, we propose TR representation of multiple image features and show that the TR representation offers better performance than the original features. Second, the integrated features by PSC is very compact and effective with lower complexity than by the standard sparse coding. Finally, the two-stage ranking strategy can balance the efficiency and memory usage in storage. Experiments demonstrate that our compact image representation is superior to the state-of-the-art alternatives for large-scale image retrieval.

    DOI

    Scopus

  • CONSTANT-TIME BILATERAL FILTER USING SPECTRAL DECOMPOSITION

    Kenjiro Sugimoto, Toby Breckon, Sei-ichiro Kamata

    2016 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     3319 - 3323  2016年  [査読有り]

     概要を見る

    This paper presents an efficient constant-time bilateral filter where constant-time means that computational complexity is independent of filter window size. Many state-of-the-art constant-time methods approximate the original bilateral filter by an appropriate combination of a series of convolutions. It is important for this framework to optimize the performance tradeoff between approximate accuracy and the number of convolutions. The proposed method achieves the optimal performance tradeoff in a least-squares manner by using spectral decomposition under the assumption that images consist of discrete intensities such as 8-bit images. This approach is essentially applicable to arbitrary range kernel. Experiments show that the proposed method outperforms state-of-the-art methods in terms of both computational complexity and approximate accuracy.

  • A NOVEL COLOR SPACE BASED ON RGB COLOR BARYCENTER

    Qieshi Zhang, Sei-ichiro Kamata

    2016 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING PROCEEDINGS   2016-May   1601 - 1605  2016年  [査読有り]

     概要を見る

    Color space is one of the bases in the image processing area. Suitable color space can give the suitable description of colors for variant processing. However, in the image processing area, the existing color space cannot show the suitable distribution in color and lightness. In this paper, a novel color space based on RGB color barycenter (RGB-CB) is proposed to describe the color and lightness more intuitively. To prove the effectiveness of the proposed color space, YUV, HSV, L* a* b*, and IPT color spaces are discussed and compared. Experimental results show the proposed color space can perform better effect than other color space in image processing.

    DOI

    Scopus

    6
    被引用数
    (Scopus)
  • LEARNING DISCRIMINATIVE AND SHAREABLE PATCHES FOR SCENE CLASSIFICATION

    Shoucheng Ni, Qieshi Zhang, Sei-ichiro Kamata, Chongyang Zhang

    2016 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING PROCEEDINGS   2016-May   1317 - 1321  2016年  [査読有り]

     概要を見る

    This paper addresses the problem of scene classification and proposes learning discriminative and shareable patches (LDSP) method. The main idea of learning discriminative and shareable patches is to discover patches that exhibit both large between-class dissimilarity (discriminative) and large within-class similarity (shareable). A novel and efficient re-clustering, based on co-occurrence relationship of first-step clustering, is proposed and conducted to further enhance the visual similarity of patches within each cluster. In order to establish appropriate criteria for selecting desired patches, a condensed representation of image features called feature epitome is introduced. In the classification, a patch feature involving pre-trained convolutional neural network model is investigated. The experimental result outperforms existing single-feature methods on MIT 67 scene benchmark in term of mean Accuracy Precision.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • EFFICIENT KEYPOINT DETECTION AND DESCRIPTION VIA POLYNOMIAL REGRESSION OF SCALE SPACE

    Ryo Okutani, Kenjiro Sugimoto, Sei-ichiro Kamata

    2016 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING PROCEEDINGS   2016-May   1357 - 1361  2016年  [査読有り]

     概要を見る

    Keypoint detection and description using approximate continuous scale space are more efficient techniques than typical discretized scale space for achieving more robust feature matching. However, this state-of-the-art method requires high computational complexity to approximately reconstruct, or decompress, the value at an arbitrary point in scale space. Specifically, it has O(M-2) computational complexity where M is an approximation order. This paper presents an efficient scale space approach that provides decompression operation with O(M) complexity without a loss of accuracy. As a result of the fact that the proposed method has much fewer variables to be solved, the least-square solution can be obtained through normal equation. This is easier to solve than the existing method which employs Karhunen-Loeve expansion and generalized eigenvalue problem. Experiments revealed that the proposed method performs as expected from the theoretical analysis.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • ADAPTIVE SAMPLING AND WAVELET TREE BASED COMPRESSIVE SENSING FOR MRI RECONSTRUCTION

    Qieshi Zhang, Jun Zhang, Sei-ichiro Kamata

    2016 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     2524 - 2528  2016年  [査読有り]

     概要を見る

    Magnetic Resonance Imaging (MRI) has been widely used in medical diagnose because of its non-invasive manner and excellent depiction of soft-tissue changes. Recently, the compressive sensing (CS) theory has been applied to reconstruct the MR image from highly down-sampled k-space data, which can reduce the scanning duration. To obtain useful information as much as possible with the same sampling rate, a weighted sampling strategy is studied. Moreover, based on the advantage of CS, a Wavelet tree based reconstruction approach is proposed. The experimental results demonstrate that the proposed method is preferable to other methods.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • A JOINTLY LOCAL STRUCTURED SPARSE DEEP LEARNING NETWORK FOR FACE RECOGNITION

    Renjie Wu, Sei-ichiro Kamata

    2016 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     3026 - 3030  2016年  [査読有り]

     概要を見る

    In this paper, we proposed an optimized Sparse Deep Learning Network (SDLN) model for Face Recognition (FR). A key contribution of this work is to learn feature coding of human face with a SDLN based on local structured Sparse Representation (SR). In traditional sparse FR methods, different poses and expressions of training samples could have great influence on the recognition results. We consider the SR that should be guided by context constraints which are defined by the correlations of dictionary atoms. The over complete common dictionary that contains common atom set has been learned from a local region structured sparse encoding process. We obtained over-complete common dictionary and feature coding for each face. As we all know that the deep learning has been widely applied to face feature learning. Using traditional deep learning methods can not contain variations of face identity information. We have to get face features of compatible change in a jointly deep learning network. The proposed SDLN is jointly fine-tuned to optimize for the task of FR. The SDLN achieves high FR performance on the ORL and FERET database.

    DOI

    Scopus

    6
    被引用数
    (Scopus)
  • EFFICIENT KEYPOINT DETECTION AND DESCRIPTION USING FILTER KERNEL DECOMPOSITION IN SCALE SPACE

    Ryo Okutani, Kenjiro Sugimoto, Sei-ichiro Kamata

    2016 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     31 - 35  2016年  [査読有り]

     概要を見る

    Keypoint detection and description in a continuous scale space achieves better robustness to scale change than those in a discretized scale space. State-of-the-art methods first decompose a continuous scale space into M + 1 component images weighted by M-order polynomials of scale sigma, and then reconstruct the value at an arbitrary point in the scale space by a linear combination of the component images. However, these methods create the mismatch that, if sigma is large, common filter kernels such as Gaussian converge to zero; but the polynomials of sigma diverge. This paper presents a more efficient approximation that suppresses this mismatch by normalizing the weighting functions. Experiments show that the proposed method achieves higher performance trade-off than state-of-the-art methods: it reduces the number of component images and total running time by 20-40% without a sacrifice of stability in keypoints detection.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Fast Bilateral Filter for Multichannel Images via Soft-assignment Coding

    Kenjiro Sugimoto, Norishige Fukushima, Sei-ichiro Kamata

    2016 ASIA-PACIFIC SIGNAL AND INFORMATION PROCESSING ASSOCIATION ANNUAL SUMMIT AND CONFERENCE (APSIPA)    2016年  [査読有り]

     概要を見る

    This paper presents an acceleration method of the bilateral filter (BF) for multi-channel images. In most existing acceleration methods, the BF is approximated by an appropriate combination of convolutions. A major purpose under this framework is to achieve sufficient approximate accuracy by as few convolutions as possible. However, state-of-the-art methods for multi-channel images still requires hundreds of (e.g., 256) convolutions to achieve sufficient accuracy. The proposed method reduces the number of convolutions without a loss in accuracy via soft-assignment coding. This approach enables us to take two major advantages that two state-of-the-art methods (scalar quantization with linear interpolation and vector quantization) have individually provided. Experiments show that the proposed method can produce sufficiently-accurate resulting images by using 64-80 convolutions only.

    DOI

    Scopus

    19
    被引用数
    (Scopus)
  • Accurate system for automatic pill recognition using imprint information

    Jiye Yu, Zhiyuan Chen, Sei-ichiro Kamata, Jie Yang

    IET IMAGE PROCESSING   9 ( 12 ) 1039 - 1047  2015年12月  [査読有り]

     概要を見る

    With rapidly advancing of contemporary medicine, it is necessary to help people identify various kinds of pills to prevent the adverse pill events. In this study, a high-accuracy automatic pill recognition system is proposed for accurate and automatic pill recognition. As pill imprint is main distinction between different pills, this system proposes algorithms on both imprint extraction and description parts to make use of imprint information. First, proposed modified stroke width transform is adopted to extract the imprint by detecting coherent strokes of imprint on the pill. Moreover, image segmentation by Loopy belief propagation is also added on printed imprint pills to solve the incoherent and coarse stroke problem. Second, a new descriptor named two-step sampling distance sets is proposed for accurate imprint description and successfully cut down the noise on extracted imprint. This strategy is based on the imprint partition - partitions the imprint on the basis of separated strokes, fragments and noise points. Recognition experiments are applied on extensive databases and result shows 90.46% rank-1 matching accuracy and 97.16% on top five ranks when classifying 12 500 query pill images into 2500 categories.

    DOI

    Scopus

    24
    被引用数
    (Scopus)
  • Compressive Bilateral Filtering

    Kenjiro Sugimoto, Sei-Ichiro Kamata

    IEEE TRANSACTIONS ON IMAGE PROCESSING   24 ( 11 ) 3357 - 3369  2015年11月  [査読有り]

     概要を見る

    This paper presents an efficient constant-time bilateral filter that produces a near-optimal performance tradeoff between approximate accuracy and computational complexity without any complicated parameter adjustment, called a compressive bilateral filter (CBLF). The constant-time means that the computational complexity is independent of its filter window size. Although many existing constant-time bilateral filters have been proposed step-by-step to pursue a more efficient performance tradeoff, they have less focused on the optimal tradeoff for their own frameworks. It is important to discuss this question, because it can reveal whether or not a constant-time algorithm still has plenty room for improvements of performance tradeoff. This paper tackles the question from a viewpoint of compressibility and highlights the fact that state-of-the-art algorithms have not yet touched the optimal tradeoff. The CBLF achieves a near-optimal performance tradeoff by two key ideas: 1) an approximate Gaussian range kernel through Fourier analysis and 2) a period length optimization. Experiments demonstrate that the CBLF significantly outperforms state-of-the-art algorithms in terms of approximate accuracy, computational complexity, and usability.

    DOI

    Scopus

    82
    被引用数
    (Scopus)
  • Efficient Constant-time Gaussian Filtering with Sliding DCT/DST-5and Dual-domain Error Minimization

    Kenjiro Sugimoto, Sei-Ichiro Kamata

    ITE Transactions on Media Technology and Applications   3 ( 1 ) 12 - 21  2015年  [査読有り]

     概要を見る

    This paper presents an efficient constant-time algorithm for Gaussian filtering and also Gaussian derivative filtering that provides a high approximate accuracy in a low computational complexity regardless of its filter window size. The proposed algorithm consists of two key techniques: second-order shift properties of the Discrete Cosine/Sine Transforms type-5 and dual-domain error minimization for finding optimal parameters. The former enables us to perform filtering in fewer number of arithmetic operations as compared than some state-of-the-art algorithms without integral images. The latter enables us to find the optimal filter size that provides the most accurate filter kernel approximation. Experiments show that the proposed algorithm clearly outperforms state-of-the-art ones in computational complexity, approximate accuracy, and accuracy stability.

    DOI

    Scopus

    30
    被引用数
    (Scopus)
  • Multi-Histogram Mapping and Fusion based Image Contrast Enhancement

    Qieshi ZHANG, Sei-ichiro KAMATA

    ITE Transactions on Media Technology and Applications   3 ( 1 ) 2 - 11  2015年01月

     概要を見る

    © 2015 by ITE Transactions on Media Technology and Applications (MTA). In this paper, a contrast enhancement method called Multi-Histogram Mapping and Fusion (MHMF) is proposed for color images. Histogram analysis based method has been successfully applied to contrast enhancement in some applications, but they are hard to enhance the dark and bright regions simultaneously for back-light images. To solve this problem, the color barycenter model (CBM) is extended to separate the color image into lightness and chroma components. Then multi-histogram mapping (MHM) is used to map the lightness component of one single color image into several new lightness components with different contrast. These new components are divided into several patches and the best patches are selected by calculating image entropy. Finally, the selected patches are fused to create the enhanced image, and mix-Gaussian filter is applied to remove the sharp transition. The experimental results show the effectiveness of proposed method comparing with other state-of-the-art methods.

    DOI CiNii

    Scopus

  • Efficient Constant-time Gaussian Filtering with Sliding DCT/DST-5and Dual-domain Error Minimization

    Kenjiro Sugimoto, Sei-Ichiro Kamata

    ITE Transactions on Media Technology and Applications   3 ( 1 ) 12 - 21  2015年

     概要を見る

    This paper presents an efficient constant-time algorithm for Gaussian filtering and also Gaussian derivative filtering that provides a high approximate accuracy in a low computational complexity regardless of its filter window size. The proposed algorithm consists of two key techniques: second-order shift properties of the Discrete Cosine/Sine Transforms type-5 and dual-domain error minimization for finding optimal parameters. The former enables us to perform filtering in fewer number of arithmetic operations as compared than some state-of-the-art algorithms without integral images. The latter enables us to find the optimal filter size that provides the most accurate filter kernel approximation. Experiments show that the proposed algorithm clearly outperforms state-of-the-art ones in computational complexity, approximate accuracy, and accuracy stability.

    DOI

    Scopus

    30
    被引用数
    (Scopus)
  • Optimized Curvelet-based Empirical Mode Decomposition

    Renjie Wu, Qieshi Zhang, Sei-ichiro Kamata

    SEVENTH INTERNATIONAL CONFERENCE ON MACHINE VISION (ICMV 2014)   9445  2015年  [査読有り]

     概要を見る

    The recent years has seen immense improvement in the development of signal processing based on Curvelet transform. The Curvelet transform provide a new multi-resolution representation. The frame elements of Curvelets exhibit higher direction sensitivity and anisotropic than the Wavelets, multi-Wavelets, steerable pyramids, and so on. These features are based on the anisotropic notion of scaling. In practical instances, time series signals processing problem is often encountered. To solve this problem, the time-frequency analysis based methods are studied. However, the time-frequency analysis cannot always be trusted. Many of the new methods were proposed. The Empirical Mode Decomposition (EMD) is one of them, and widely used. The EMD aims to decompose into their building blocks functions that are the superposition of a reasonably small number of components, well separated in the time-frequency plane. And each component can be viewed as locally approximately harmonic. However, it cannot solve the problem of directionality of high-dimensional. A reallocated method of Curvelet transform (optimized Curvelet-based EMD) is proposed in this paper. We introduce a definition for a class of functions that can be viewed as a superposition of a reasonably small number of approximately harmonic components by optimized Curvelet family. We analyze this algorithm and demonstrate its results on data. The experimental results prove the effectiveness of our method.

    DOI

    Scopus

  • Disparity Estimation from Monocular Image Sequence

    Qieshi Zhang, Sei-ichiro Kamata

    SEVENTH INTERNATIONAL CONFERENCE ON MACHINE VISION (ICMV 2014)   9445  2015年  [査読有り]

     概要を見る

    This paper proposes a novel method for estimating disparity accurately. To achieve the ideal result, an optimal adjusting framework is proposed to address the noise, occlusions, and outliners. Different from the typical multi-view stereo (MVS) methods, the proposed approach not only use the color constraint, but also use the geometric constraint associating multiple frame from the image sequence. The result shows the disparity with a good visual quality that most of the noise is eliminated, the errors in occlusion area are suppressed and the details of scene objects are preserved.

    DOI

    Scopus

  • Sparse Decomposition Learning Based Dynamic MRI Reconstruction

    Peifei Zhu, Qieshi Zhang, Sei-ichiro Kamata

    SEVENTH INTERNATIONAL CONFERENCE ON MACHINE VISION (ICMV 2014)   9445  2015年  [査読有り]

     概要を見る

    Dynamic MRI is widely used for many clinical exams but slow data acquisition becomes a serious problem. The application of Compressed Sensing (CS) demonstrated great potential to increase imaging speed. However, the performance of CS is largely depending on the sparsity of image sequence in the transform domain, where there are still a lot to be improved. In this work, the sparsity is exploited by proposed Sparse Decomposition Learning (SDL) algorithm, which is a combination of low-rank plus sparsity and Blind Compressed Sensing (BCS). With this decomposition, only sparsity component is modeled as a sparse linear combination of temporal basis functions. This enables coefficients to be sparser and remain more details of dynamic components comparing learning the whole images. A reconstruction is performed on the undersampled data where joint multicoil data consistency is enforced by combing Parallel Imaging (PI). The experimental results show the proposed methods decrease about 15 similar to 20% of Mean Square Error (MSE) compared to other existing methods.

    DOI

    Scopus

  • Disparity Refinement with Stability-based Tree for Stereo Matching

    Yuhang Ji, Qieshi Zhang, Kenjiro Sugimoto, Sei-ichiro Kamata

    2015 IEEE INTELLIGENT VEHICLES SYMPOSIUM (IV)     469 - 474  2015年  [査読有り]

     概要を見る

    This paper proposes a disparity refinement method with stability-based tree. By developing stability-based tree to evaluate and reconstruct support regions for error parts, the proposed method achieves effective performance in removing outliers. This approach further improves the quality of raw disparity map in stereo matching, which makes the local methods results comparable to the global ones. Experiments exhibit that the proposed method reduces more than 70% aggregation time compared with traditional tree method without loss of accuracy. It also outperforms existing disparity refinement methods in removing large error parts.

  • Autonomous driving experiments by Small electric vehicle in simulated road

    Nan Wu, Qieshi Zhang, Xun Pan, Hu Beier, Harutoshi Ogai, Sei-ichiro Kamata, Hiroshi Inujima, Shigeyuki Tateno

    2015 54TH ANNUAL CONFERENCE OF THE SOCIETY OF INSTRUMENT AND CONTROL ENGINEERS OF JAPAN (SICE)     1451 - 1452  2015年  [査読有り]

     概要を見る

    © 2015 The Society of Instrument and Control Engineers-SICE. In the city of Kitakyushu, a quarter more of people are older than 65 years old. Especially a large proportion of elderly people are living on their own. So how to let old people easily access the communal facilities like hospitals need be attended. Therefore, the bad connection between public transportation and home is need to be solved. Based on the survey result and road situation in Kitakyushu city, a new type small single-seat electrical vehicle (sEV) is studied to provide a solution for elder to easily access the public transportation. In this research, low cost and safe automatic driving electrical vehicle based on limited number of sensors is focused.

    DOI

    Scopus

  • Robust Road Lane Detection using Extremal-Region Enhancement

    Jingchen Gu, Qieshi Zhang, Sei-ichiro Kamata

    Proceedings 3rd IAPR Asian Conference on Pattern Recognition ACPR 2015     519 - 523  2015年  [査読有り]

     概要を見る

    Road lane detection is a key problem in advanced driver-assistance systems (ADAS). For solving this problem, vision-based detection methods are widely used and are generally focused on edge information. However, only using edge information leads to miss detection and error detection in various road conditions. In this paper, we propose a neighbor-based image conversion method, called extremal-region enhancement. The proposed method enhances the white lines in intensity, hence it is robust to shadows and illuminance changes. Both edge and shape information of white lines are extracted as lane features in the method. In addition, we implement a robust road lane detection algorithm using the extracted features and improve the correctness through probability tracking. The experimental result shows an average detection rate increase of 13.2% over existing works.

    DOI

    Scopus

    8
    被引用数
    (Scopus)
  • Fisheye Image Correction Based on Straight-line Detection and Preservation

    Qieshi Zhang, Sei-ichiro Kamata

    2015 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN, AND CYBERNETICS (SMC 2015): BIG DATA ANALYTICS FOR HUMAN-CENTRIC SYSTEMS     1793 - 1797  2015年  [査読有り]

     概要を見る

    Fisheye lenses are widely used when the users want to capture the image/video with wide field of view (FoV) which is particularly suited to surveillance monitoring and vehicle camera. However, no projection from the actual scene in wide FoV image can avoid the distortion. If this problem cannot be solved, the fisheye image will difficult be used for object detection or analysis due to the distorted shapes of the scene objects. To correct this problem and obtain the natural-looking image, a two-step correction approach is proposed. Firstly, adaptive latitude and longitude correction are presented and the Hough transform is used to detect and estimate the straight-line. Secondly, the straight-line preserving and orientation, consistency based optimization is examined to obtain the final correction result. To compare the effectiveness of the proposed method, some fisheye correction methods are discussed. The experimental results demonstrate that the proposed method can obtain the coherent natural-looking.

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • Pill Recognition Using Imprint Information by Two-step Sampling Distance Sets

    Jiye Yu, Zhiyuan Chen, Sei-ichiro Kamata

    2014 22ND INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR)     3156 - 3161  2014年  [査読有り]

     概要を見る

    Huge variety of medicine cures diseases. But unlabeled pills sometimes confuse people, even causing adverse drug events. This paper introduces a high accuracy automatic pill recognition method based on pill imprint which is a main discriminative factor between different pills. To describe the imprint information clearly, we propose a Two-step Sampling Distance Sets (TSDS) descriptor based on Distance Sets (DS) using a two-step sampling strategy. The two-step sampling strategy applies a resampling according to imprint segmentation, which divides an imprint into separated strokes, fragments and noise points. The TSDS is able to take control over the selection of feature points, aiming to cut down the noise points and unwished fragments generated by imprint extraction which will cause disturbance on recognition. In the aspect of the imprint extraction, we preprocess the pill image by dynamic contrast adjustment to cope with the exposure problem. Modified Stroke Width Transform (MSWT) is used to extract the imprint by detecting the coherent strokes on the pill. Finally, several experimental results have shown 86.01%, rank-1 matching accuracy, and 93.64%, within top 5 ranks, when classifying pills into 2500 categories.

    DOI

    Scopus

    18
    被引用数
    (Scopus)
  • Development of autonomous small EV in Japan aging society

    Nan Wu, Harutoshi Ogai, Masakuni Ohshiro, Seiichirou Kamata, Shigeyuki Tateno, Akira Uchida, Masahiko Kai, Makio Iida, Yuji Sano

    IFAC Proceedings Volumes (IFAC-PapersOnline)   3 ( 1 ) 966 - 972  2014年

     概要を見る

    In Kitakyushu city, more than 25 % people are older than 65 years old. The roads in residential area for the people are very narrow, very steep slope and vulnerable. Some needs assessment for small electric vehicle at some event and community activity was done. Based on the survey result and topographical features, the small electric vehicle for elderly person was selected and was modified and the automatic driving system was built. Automatic driving and platooning using Zigbee or Digimesh to exchange driving data between vehicles were built and tested. © 2014 IFAC.

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • O(1) Transposed Bilateral Filtering for Optimization

    Kenjiro Sugimoto, Keiichiro Shirai, Sei-ichiro Kamata

    2014 ASIA-PACIFIC SIGNAL AND INFORMATION PROCESSING ASSOCIATION ANNUAL SUMMIT AND CONFERENCE (APSIPA)    2014年  [査読有り]

     概要を見る

    This paper presents an essential algorithm for optimization-based image processing using the bilateral filter (BLF), called constant-time transposed BLF (O(1) TBLF). Some iterative solvers for optimization problems require a pair of filters defined as multiplying a filter matrix or its transpose to vectorized images. Since the BLF can be described as a matrix form, its paired filter also exists, called a TBLF in this paper. BLF-based optimization achieves high smoothing performance; whereas, it requires much high computational complexity due to iterating both BLF and TBLF many times. Hence, this paper designs an O(1) TBLF algorithm to accelerate the iterative process. Experiments show that our O(1) TBLF runs in low complexity regardless of its filter window size and works effectively for flash/no-flash image integration via BLF-based optimization.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Maximum correntropy criterion for discriminative dictionary learning

    Hao, Pengyi, Kamata, Sei Ichiro

    2013 IEEE International Conference on Image Processing, ICIP 2013 - Proceedings     4325 - 4329  2013年12月  [査読有り]

     概要を見る

    In this paper, a novel discriminative dictionary learning with pairwise constraints by maximum correntropy criterion is proposed for pair matching problem. Comparing with the conventional dictionary learning approaches, the proposed method has several advantages: (i) It can deal with the outliers and noises problem more efficiently during the reconstruction step. (ii) It can be effectively solved by half-quadratic optimization algorithm, and in each iteration step, the complex optimization problem can be reduced to a general problem that can be efficiently solved by feature-sign search optimization. (iii) The proposed method is capable of analyzing non-Gaussian noise to reduce the influence of large outliers substantially, resulting in a robust and discriminative dictionary. We test the performance of the proposed method on two applications: face verification on the challenging restricted protocol of Labeled Faces in the Wild (LFW) benchmark and face-track identification on a dataset with more than 7,000 face-tracks. Compared with the recent state-of-the-art approaches, the outstanding performance of the proposed method validates its robustness and discriminability. © 2013 IEEE.

    DOI

    Scopus

    6
    被引用数
    (Scopus)
  • Improved Color Barycenter Model and Its Separation for Road Sign Detection

    Qieshi Zhang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E96D ( 12 ) 2839 - 2849  2013年12月  [査読有り]

     概要を見る

    This paper proposes an improved color barycenter model (CBM) and its separation for automatic road sign (RS) detection. The previous version of CBM can find out the colors of RS, but the accuracy is not high enough for separating the magenta and blue regions and the influence of number with the same color are not considered. In this paper, the improved CBM expands the barycenter distribution to cylindrical coordinate system (CCS) and takes the number of colors at each position into account for clustering. Under this distribution, the color information can be represented more clearly for analyzing. Then aim to the characteristic of barycenter distribution in CBM (CBM-BD), a constrained clustering method is presented to cluster the CBM-BD in CCS. Although the proposed clustering method looks like conventional K-means in some part, it can solve some limitations of K-means in our research. The experimental results show that the proposed method is able to detect RS with high robustness.

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • A Novel Color Descriptor for Road-Sign Detection

    Qieshi Zhang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES   E96A ( 5 ) 971 - 979  2013年05月  [査読有り]

     概要を見る

    This paper presents a novel color descriptor based on the proposed Color Barycenter Hexagon (CBH) model for automatic Road-Sign (RS) detection. In the visual Driver Assistance System (DAS), RS detection is one of the most important factors. The system provides drivers with important information on driving safety. Different color combinations of RS indicate different functionalities; hence a robust color detector should be designed to address color changes in natural surroundings. The CBH model is constructed with barycenter distribution in the created color triangle, which represents RS colors in a more compact way. For detecting RS, the CBH model is used to segment color information at the initial step. Furthermore, a judgment process is applied to verify each RS candidate through the size, aspect ratio, and color ratio. Experimental results show that the proposed method is able to detect RS with robust, accurate performance and is invariant to light and scale in more complex surroundings.

    DOI

    Scopus

    9
    被引用数
    (Scopus)
  • L1-Norm Based Linear Discriminant Analysis: An Application to Face Recognition

    Wei Zhou, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E96D ( 3 ) 550 - 558  2013年03月  [査読有り]

     概要を見る

    Linear Discriminant Analysis (LDA) is a well-known feature extraction method for supervised subspace learning in statistical pattern recognition. In this paper, a novel method of LDA based on a new L1-norm optimization technique and its variances are proposed. The conventional LDA, which is based on L2-norm, is sensitivity to the presence of outliers, since it used the L2-norm to measure the between-class and within-class distances. In addition, the conventional LDA often suffers from the so-called small sample size (3S) problem since the number of samples is always smaller than the dimension of the feature space in many applications, such as face recognition. Based on L1-norm, the proposed methods have several advantages, first they are robust to outliers because they utilize the L1-norm, which is less sensitive to outliers. Second, they have no 3S problem. Third, they are invariant to rotations as well. The proposed methods are capable of reducing the influence of outliers substantially, resulting in a robust classification. Performance assessment in face application shows that the proposed approaches are more effectiveness to address outliers issue than traditional ones.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Development of safety assist system for ultra-small EV - Efforts for safe mobility in Kitakyushu aging society

    Ogai, Harutoshi, Kamata, Seiichirou, Wu, Nan, Ishi, Taro, Uchida, Akira, Kai, Masahiko, Iida, Makio

    20th ITS World Congress Tokyo 2013    2013年01月

     概要を見る

    In Kitakyushu city there are more than a quarter of people are older than 65-year-old. The road in residential area of this area has specific characters including very narrow road, very steep slope and vulnerable roadbed. Some needs assessment of small electric vehicle at some event activity and community was done. Based on the survey result and topographical features, the small electric vehicle for elder was designed and modified and automatic driving system was built. Automatic driving and platooning using Zigbee to exchange driving data between vehicles were built and tested.

  • A new accurate pill recognition system using imprint information

    Zhiyuan Chen, Sei-Ichiro Kamata

    Proceedings of SPIE - The International Society for Optical Engineering   9067  2013年

     概要を見る

    Great achievements in modern medicine benefit human beings. Also, it has brought about an explosive growth of pharmaceuticals that current in the market. In daily life, pharmaceuticals sometimes confuse people when they are found unlabeled. In this paper, we propose an automatic pill recognition technique to solve this problem. It functions mainly based on the imprint feature of the pills, which is extracted by proposed MSWT (modified stroke width transform) and described by WSC (weighted shape context). Experiments show that our proposed pill recognition method can reach an accurate rate up to 92.03% within top 5 ranks when trying to classify more than 10 thousand query pill images into around 2000 categories. © 2013 SPIE.

    DOI

    Scopus

    6
    被引用数
    (Scopus)
  • Face recognition with learned local curvelet patterns and 2-directional L1-norm based 2DPCA

    Wei Zhou, Sei-Ichiro Kamata

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)   7728 ( 1 ) 109 - 120  2013年

     概要を見る

    In this paper, we propose Learned Local Curvelet Patterns (LLCP) for presenting the local features of facial images. The proposed method is based on curvelet transform which can overcome the weakness of traditional Gabor wavelets in higher dimension, and better capture the curve singularities and hyperplane singularities of facial images. Different from wavelet transform, curvelet transform can effectively and efficiently approximate the curved edges with very few coefficients as well as taking space-frequency information into consideration. First, LLCP designs several learned codebooks from Curvelet filtered facial images. Then each facial image can be encoded into multiple pattern maps and finally block-based histograms of these patterns are concatenated into an histogram sequence to be used as a face descriptor. In order to reduce the face feature descriptor, 2-Directional L1-Norm Based 2DPCA ((2D)2PCA-L1) is proposed which is simultaneously considering the row and column directions for efficient face representation and recognition. Performance assessment in several face recognition problem shows that the proposed approach is superior to traditional ones. © 2013 Springer-Verlag.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Linear discriminant analysis with maximum correntropy criterion

    Wei Zhou, Sei-Ichiro Kamata

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)   7724 ( 1 ) 500 - 511  2013年

     概要を見る

    Linear Discriminant Analysis (LDA) is a famous supervised feature extraction method for subspace learning in computer vision and pattern recognition. In this paper, a novel method of LDA based on a new Maximum Correntropy Criterion optimization technique is proposed. The conventional LDA, which is based on L2-norm, is sensitivity to the presence of outliers. The proposed method has several advantages: first, it is robust to large outliers. Second, it is invariant to rotations. Third, it can be effectively solved by half-quadratic optimization algorithm. And in each iteration step, the complex optimization problem can be reduced to a quadratic problem that can be efficiently solved by a weighted eigenvalue optimization method. The proposed method is capable of analyzing non-Gaussian noise to reduce the influence of large outliers substantially, resulting in a robust classification. Performance assessment in several datasets shows that the proposed approach is more effectiveness to address outlier issue than traditional ones. © 2013 Springer-Verlag.

    DOI

    Scopus

    4
    被引用数
    (Scopus)
  • Facial signatures for fast individual retrieval from video dataset

    Pengyi Hao, Sei-Ichiro Kamata

    Proceedings - IEEE International Conference on Multimedia and Expo     1 - 6  2013年

     概要を見る

    The topic of retrieving videos containing a desired person by using the content of faces without any help of textual information has many interesting applications like video surveillance, social network, video mining, etc. However, face-by-face matching leads to an unacceptable response time for a video dataset with a large number of detected faces and may also reduce the accuracy of searching. Therefore, in this paper we propose a scheme to generate facial signatures for fast retrieving videos containing the same person with a query. First, we summarize each video as a set of person-oriented individuals based on detected faces, which are represented as high dimensional vectors in a feature space. Then, each person with a collection of high dimensional vectors is projected to a compact and reduced dimensionality representation that is called facial signature for this person. The projection is realized by constructing a matcher using linear discriminant analysis with maximum correntropy criterion optimization. In this research, two kinds of signatures are provided, which are called 1D facial signature and 2D facial signature. The proposed searching scheme can support two types of queries: face image and video clip. Evaluations on a large dataset of videos show reliable measurement of similarities by using facial signature to represent each person generated from videos and also demonstrate that the proposed searching scheme has the potential to substantially reduce the response time and slightly increase the mean average precision of retrieval. © 2013 IEEE.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Efficient large-scale video retrieval via discriminative signatures

    Pengyi Hao, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E96-D ( 8 ) 1800 - 1810  2013年

     概要を見る

    The topic of retrieving videos containing a desired person from a dataset just using the content of faces without any help of textual information has many interesting applications like video surveillance, social network, video mining, etc. However, traditional face matching against a huge number of detected faces leads to an unacceptable response time and may also reduce the accuracy due to the large variations in facial expressions, poses, lighting, etc. Therefore, in this paper we propose a novel method to generate discriminative "signatures" for efficiently retrieving the videos containing the same person with a query. In this research, the signature is defined as a compact, discriminative and reduced dimensionality representation, which is generated from a set of high-dimensional feature vectors of an individual. The desired videos are retrieved based on the similarities between the signature of the query and those of individuals in the database. In particular, we make the following contributions. Firstly, we give an algorithm of two directional linear discriminant analysis with maximum correntropy criterion (2DLDA-MCC) as an extension to our recently proposed maximum correntropy criterion based linear discriminant analysis (LDA-MCC). Both algorithms are robust to outliers and noise. Secondly, we present an approach for transferring a set of exemplars to a fixed-length signature using LDA-MCC and 2DLDA-MCC, resulting in two kinds of signatures that are called 1D signature and 2D signature. Finally, a novel video retrieval scheme is given based on the signatures, which has low storage requirement and can achieve a fast search. Evaluations on a large dataset of videos show reliable measurement of similarities by using the proposed signatures to represent the identities generated from videos. Experimental results also demonstrate that the proposed video retrieval scheme has the potential to substantially reduce the response time and slightly increase the mean average precision of retrieval. Copyright © 2013 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • AN EFFICIENT VIDEO RETRIEVAL SCHEME BASED ON FACIAL SIGNATURES

    Pengyi Hao, Sei-ichiro Kamata

    2013 20TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2013)     2699 - 2703  2013年  [査読有り]

     概要を見る

    The topic of retrieving videos containing a desired person by just using facial content has many applications like video surveillance, social network, etc. In this paper, we propose a compact, discriminative and low-dimensional signature to describe an person with a set of high-dimensional features. The signature is generated by linear discriminant analysis with maximum correntropy criterion that is robust to outliers and noises. Based on the proposed signatures, a new video retrieval scheme is given for fast finding the desired videos by measuring the similarities between the signature of a query and the ones in the dataset. Evaluations on a large dataset of videos show that the proposed video retrieval scheme has the potential to substantially reduce the response time and slightly increase the mean average precision of retrieval.

  • A FAST MULTI-VIEW BASED SPECULAR REMOVAL APPROACH FOR PILL EXTRACTION

    Chengjie Wang, Sei-ichiro Kamata, Lizhuang Ma

    2013 20TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2013)     4126 - 4130  2013年  [査読有り]

     概要を見る

    This paper presents a novel approach to remove the specular reflections on the transparent plastic medicine package and automatically extract the randomly distributed pills inside. In this approach, three cameras are employed to take images of the package from different viewpoints. And these three images are used as input image set while the output is a series of small images of a single pill. And these images can be directly applied to the traditional single pill recognition algorithms. The experimental results show the reliability of our approach by measuring correct detection rate (100%), false detection rate (0%) and pill separation accuracy (98.4%). And the proposed method processes a set of three 725 x 725 sized images at 0.15s averagely on a Core i5-2400 3.1GHz PC.

  • A FOREGROUND OBJECT BASED QUANTITATIVE ASSESSMENT OF DENSE STEREO APPROACHES FOR USE IN AUTOMOTIVE ENVIRONMENTS

    Oliver K. Hamilton, Toby P. Breckon, Xuejiao Bai, Sei-ichiro Kamata

    2013 20TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2013)     418 - 422  2013年  [査読有り]

     概要を見る

    There has been significant recent interest in stereo correspondence algorithms for use in the urban automotive environment [1, 2, 3]. In this paper we evaluate a range of dense stereo algorithms, using a unique evaluation criterion which provides quantitative analysis of accuracy against range, based on ground truth 3D annotated object information. The results show that while some algorithms provide greater scene coverage, we see little differentiation in accuracy over short ranges, while the converse is shown over longer ranges. Within our long range accuracy analysis we see a distinct separation of relative algorithm performance. This study extends prior work on dense stereo evaluation of Block Matching (BM)[4], Semi-Global Block Matching (SGBM)[5], No Maximal Disparity (NoMD)[6], Cross[7], Adaptive Dynamic Programming (AdptDP)[8], Efficient Large Scale (ELAS)[9], Minimum Spanning Forest (MSF)[10] and Non-Local Aggregation (NLA)[11] using a novel quantitative metric relative to object range.

  • FAST GAUSSIAN FILTER WITH SECOND-ORDER SHIFT PROPERTY OF DCT-5

    Kenjiro Sugimoto, Sei-ichiro Kamata

    2013 20TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2013)     514 - 518  2013年  [査読有り]

     概要を見る

    This paper presents an efficient constant-time Gaussian filter which provides a high accuracy at a low cost over a wide range of scale sigma. It requires only 14 multiplications per pixel in image filtering regardless of sigma, which is fewer than state-of-the- art constant-time Gaussian filters. Main ideas of the paper are as follows: 1) introducing a second-order shift property of the discrete cosine transform type-5 (DCT-5) to convolve cosines faster, and 2) suppressing error propagation caused by the shift property. Experiments in image processing show that the proposed algorithm is 3.7 x faster than a state-of-the-art recursive Gaussian filter and comparable to that of +/- 3 sigma-ssupported Gaussian convolution with sigma = 2.33. The output accuracy is stable at around 80 [dB] all over sigma is an element of [1, 128].

  • Facial signatures for fast individual retrieval from video dataset

    Pengyi Hao, Sei-Ichiro Kamata

    Proceedings - IEEE International Conference on Multimedia and Expo    2013年

     概要を見る

    The topic of retrieving videos containing a desired person by using the content of faces without any help of textual information has many interesting applications like video surveillance, social network, video mining, etc. However, face-by-face matching leads to an unacceptable response time for a video dataset with a large number of detected faces and may also reduce the accuracy of searching. Therefore, in this paper we propose a scheme to generate facial signatures for fast retrieving videos containing the same person with a query. First, we summarize each video as a set of person-oriented individuals based on detected faces, which are represented as high dimensional vectors in a feature space. Then, each person with a collection of high dimensional vectors is projected to a compact and reduced dimensionality representation that is called facial signature for this person. The projection is realized by constructing a matcher using linear discriminant analysis with maximum correntropy criterion optimization. In this research, two kinds of signatures are provided, which are called 1D facial signature and 2D facial signature. The proposed searching scheme can support two types of queries: face image and video clip. Evaluations on a large dataset of videos show reliable measurement of similarities by using facial signature to represent each person generated from videos and also demonstrate that the proposed searching scheme has the potential to substantially reduce the response time and slightly increase the mean average precision of retrieval. © 2013 IEEE.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • AN EFFICIENT VIDEO RETRIEVAL SCHEME BASED ON FACIAL SIGNATURES

    Pengyi Hao, Sei-ichiro Kamata

    2013 20TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2013)     2699 - 2703  2013年  [査読有り]

     概要を見る

    The topic of retrieving videos containing a desired person by just using facial content has many applications like video surveillance, social network, etc. In this paper, we propose a compact, discriminative and low-dimensional signature to describe an person with a set of high-dimensional features. The signature is generated by linear discriminant analysis with maximum correntropy criterion that is robust to outliers and noises. Based on the proposed signatures, a new video retrieval scheme is given for fast finding the desired videos by measuring the similarities between the signature of a query and the ones in the dataset. Evaluations on a large dataset of videos show that the proposed video retrieval scheme has the potential to substantially reduce the response time and slightly increase the mean average precision of retrieval.

    DOI

    Scopus

  • A FAST MULTI-VIEW BASED SPECULAR REMOVAL APPROACH FOR PILL EXTRACTION

    Chengjie Wang, Sei-ichiro Kamata, Lizhuang Ma

    2013 20TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2013)     4126 - 4130  2013年  [査読有り]

     概要を見る

    This paper presents a novel approach to remove the specular reflections on the transparent plastic medicine package and automatically extract the randomly distributed pills inside. In this approach, three cameras are employed to take images of the package from different viewpoints. And these three images are used as input image set while the output is a series of small images of a single pill. And these images can be directly applied to the traditional single pill recognition algorithms. The experimental results show the reliability of our approach by measuring correct detection rate (100%), false detection rate (0%) and pill separation accuracy (98.4%). And the proposed method processes a set of three 725 x 725 sized images at 0.15s averagely on a Core i5-2400 3.1GHz PC.

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • FAST GAUSSIAN FILTER WITH SECOND-ORDER SHIFT PROPERTY OF DCT-5

    Kenjiro Sugimoto, Sei-ichiro Kamata

    2013 20TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2013)     514 - 518  2013年  [査読有り]

     概要を見る

    This paper presents an efficient constant-time Gaussian filter which provides a high accuracy at a low cost over a wide range of scale sigma. It requires only 14 multiplications per pixel in image filtering regardless of sigma, which is fewer than state-of-the- art constant-time Gaussian filters. Main ideas of the paper are as follows: 1) introducing a second-order shift property of the discrete cosine transform type-5 (DCT-5) to convolve cosines faster, and 2) suppressing error propagation caused by the shift property. Experiments in image processing show that the proposed algorithm is 3.7 x faster than a state-of-the-art recursive Gaussian filter and comparable to that of +/- 3 sigma-ssupported Gaussian convolution with sigma = 2.33. The output accuracy is stable at around 80 [dB] all over sigma is an element of [1, 128].

    DOI

    Scopus

    40
    被引用数
    (Scopus)
  • Face Representation and Recognition with Local Curvelet Patterns

    Wei Zhou, Alireza Ahrary, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E95D ( 12 ) 3078 - 3087  2012年12月  [査読有り]

     概要を見る

    In this paper, we propose Local Curve let Binary Patterns (LCBP) and Learned Local Curve let Patterns (LLCP) for presenting the local features of facial images. The proposed methods are based on Curve let transform which can overcome the weakness of traditional Gabor wavelets in higher dimensions, and better capture the curve singularities and hyperplane singularities of facial images. LCBP can be regarded as a combination of Curve let features and LBP operator while LLCP designs several learned codebooks from patch sets, which are constructed by sampling patches from Curvelet filtered facial images. Each facial image can be encoded into multiple pattern maps and block-based histograms of these patterns are concatenated into an histogram sequence to be used as a face descriptor. During the face representation phase, one input patch is encoded by one pattern in LCBP while multi-patterns in LLCP. Finally, an effective classifier called Weighted Histogram Spatially constrained Earth Mover's Distance (WHSEMD) which utilizes the discriminative powers of different facial parts, the different patterns and the spatial information of face is proposed. Performance assessment in face recognition and gender estimation under different challenges shows that the proposed approaches are superior than traditional ones.

    DOI

    Scopus

  • Individual authentication through hand posture recognition using Multi-Hilbert Scanning Distance

    Ryu, Jegoon, Kamata, Sei Ichiro

    European Signal Processing Conference     1787 - 1790  2012年11月

     概要を見る

    In this paper, we propose a novel Hand Posture Recognition (HPR) for biometrics. This study uses the three dimensional point clouds for robust hand posture recognition at the rotation and scale. Multi-Hilbert Scanning Distance (MHSD) are also introduced for mathematical approaches of shape matching. HPR framework is divided into five parts: detecting hand region, removing the wrist, aligning the hand pose, extracting feature descriptor, and matching. Based on the experimental results, this framework showed superior results for hand posture recognition rate. © 2012 EURASIP.

  • A Histogram Separation and Mapping Framework for Image Contrast Enhancement

    Qieshi Zhang, Sei-ichiro Kamata

    IPSJ Transactions on Computer Vision and Applications   4   100 - 107  2012年09月

     概要を見る

    In this paper, an adaptive framework based on histogram separation and mapping for image contrast enhancement is presented. In this framework, the histogram is separated by binary tree structure with the proposed adaptive histogram separation strategy. Generally, histogram equalization (HE) is an effective technique for contrast enhancement. However, the conventional HE usually gives the processed image with unnatural look and artifacts by excessive enhancement. For overcoming this shortage, the adaptive histogram separation unit (AHSU) is proposed to convert the global enhancement problem into local. And for mapping the histogram partitions into more optimal ranges, the exact histogram separation is discussed. Finally, an adaptive histogram separation and mapping framework (AHSMF) for contrast enhancement is presented, and the experimental results show better effectiveness than other histogram based methods.

    DOI CiNii

  • A Simple and Effective Clustering Algorithm for Multispectral Images Using Space-Filling Curves

    Jian Zhang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E95D ( 7 ) 1749 - 1757  2012年07月  [査読有り]

     概要を見る

    With the wide usage of multispectral images, a fast efficient multidimensional clustering method becomes not only meaningful but also necessary. In general, to speed up the multidimensional images' analysis, a multidimensional feature vector should be transformed into a lower dimensional space. The Hilbert curve is a continuous one-to-one mapping from N-dimensional space to one-dimensional space, and can preserves neighborhood as much as possible. However, because the Hilbert curve is generated by a recurve division process, 'Boundary Effects' will happen, which means data that are close in N-dimensional space may not be close in one-dimensional Hilbert order. In this paper, a new efficient approach based on the space-filling curves is proposed for classifying multispectral satellite images. In order to remove 'Boundary Effects' of the Hilbert curve, multiple Hilbert curves, z curves, and the Pseudo-Hilbert curve are used jointly. The proposed method extracts category clusters from one-dimensional data without computing any distance in N-dimensional space. Furthermore, multispectral images can be analyzed hierarchically from coarse data distribution to fine data distribution in accordance with different application. The experimental results performed on LANDSAT data have demonstrated that the proposed method is efficient to manage the multispectral images and can be applied easily.

    DOI

    Scopus

  • SSM-HPC: Front View Gait Recognition Using Spherical Space Model with Human Point Clouds

    Jegoon Ryu, Sei-ichiro Kamata, Alireza Ahrary

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E95D ( 7 ) 1969 - 1978  2012年07月  [査読有り]

     概要を見る

    In this paper, we propose a novel gait recognition framework - Spherical Space Model with Human Point Clouds (SSM-HPC) to recognize front view of human gait. A new gait representation - Marching in Place (MIP) gait is also introduced which preserves the spatiotemporal characteristics of individual gait manner. In comparison with the previous studies on gait recognition which usually use human silhouette images from image sequences, this research applies three dimensional (3D) point clouds data of human body obtained from stereo camera. The proposed framework exhibits gait recognition rates superior to those of other gait recognition methods.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Visibility restoration from single image based optical model

    Zhang, Qieshi, Zhang, Qieshi, Kamata, Sei Ichiro

    VISAPP 2012 - Proceedings of the International Conference on Computer Vision Theory and Applications   1   213 - 216  2012年06月

     概要を見る

    In this paper, we propose a segmentation based method to estimate the haze-free image by the optical model. In this work, we estimate the atmospheric light by color barycenter hexagon (CBH) model and use the watershed to segment the image to calculate transmission map by dark pixels with single image. Firstly, non-color region is segmented by CBH model and calculate the atmospheric light. Then, use the watershed with rang component of CBH model to segment the color image into several sub-regions, and estimate the transmission map. Finally, use the optical model with the parameters to restore the haze-free image. The experimental results show that our method is more effective and able to get better results than other compared single image based methods.

  • Efficiently Finding Individuals from Video Dataset

    Pengyi Hao, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E95D ( 5 ) 1280 - 1287  2012年05月  [査読有り]

     概要を見る

    We are interested in retrieving video shots or videos containing particular people from a video dataset. Owing to the large variations in pose, illumination conditions, occlusions, hairstyles and facial expressions, face tracks have recently been researched in the fields of face recognition, face retrieval and name labeling from videos. However, when the number of face tracks is very large, conventional methods, which match all or some pairs of faces in face tracks, will not be effective. Therefore, in this paper, an efficient method for finding a given person from a video dataset is presented. In our study, in according to performing research on face tracks in a single video, we also consider how to organize all the faces in videos in a dataset and how to improve the search quality in the query process. Different videos may include the same person; thus, the management of individuals in different videos will be useful for their retrieval. The proposed method includes the following three points. (i) Face tracks of the same person appearing for a period in each video are first connected on the basis of scene information with a time constriction, then all the people in one video are organized by a proposed hierarchical clustering method. (ii) After obtaining the organizational structure of all the people in one video, the people are organized into an upper layer by affinity propagation. (iii) Finally, in the process of querying, a remeasuring method based on the index structure of videos is performed to improve the retrieval accuracy. We also build a video dataset that contains six types of videos: films, TV shows, educational videos, interviews, press conferences and domestic activities. The formation of face tracks in the six types of videos is first researched, then experiments are performed on this video dataset containing more than 1 million faces and 218,786 face tracks. The results show that the proposed approach has high search quality and a short search time.

    DOI

    Scopus

    6
    被引用数
    (Scopus)
  • Novel Algorithm for Polar and Spherical Fourier Analysis on Two and Three Dimensional Images

    Zhuo Yang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E95D ( 5 ) 1248 - 1255  2012年05月  [査読有り]

     概要を見る

    Polar and Spherical Fourier analysis can be used to extract rotation invariant features for image retrieval and pattern recognition tasks. They are demonstrated to show superiorities comparing with other methods on describing rotation invariant features of two and three dimensional images. Based on mathematical properties of trigonometric functions and associated Legendre polynomials, fast algorithms are proposed for multimedia applications like real time systems and large multimedia databases in order to increase the computation speed. The symmetric points are computed simultaneously. Inspired by relative prime number theory, systematic analysis are given in this paper. Novel algorithm is deduced that provide even faster speed. Proposed method are 9-15% faster than previous work. The experimental results on two and three dimensional images are given to illustrate the effectiveness of the proposed method. Multimedia signal processing applications that need real time polar and spherical Fourier analysis can be benefit from this work.

    DOI

    Scopus

  • Image Description with Local Patterns: An Application to Face Recognition

    Wei Zhou, Alireza Ahrary, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E95D ( 5 ) 1494 - 1505  2012年05月

     概要を見る

    In this paper, we propose a novel approach for presenting the local features of digital image using 1D Local Patterns by Multi-Scans (1DLPMS). We also consider the extentions and simplifications of the proposed approach into facial images analysis. The proposed approach consists of three steps. At the first step, the gray values of pixels in image are represented as a vector giving the local neighborhood intensity distrubutions of the pixels. Then, multi-scans are applied to capture different spatial information on the image with advantage of less computation than other traditional ways, such as Local Binary Patterns (LBP). The second step is encoding the local features based on different encoding rules using 1D local patterns. This transformation is expected to be less sensitive to illumination variations besides preserving the appearance of images embedded in the original gray scale. At the final step, Grouped 1D Local Patterns by Multi-Scans (G1DLPMS) is applied to make the proposed approach computationally simpler and easy to extend. Next, we further formulate boosted algorithm to extract the most discriminant local features. The evaluated results demonstrate that the proposed approach outperforms the conventional approaches in terms of accuracy in applications of face recognition, gender estimation and facial expression.

    DOI

    Scopus

    10
    被引用数
    (Scopus)
  • A Linear Manifold Color Descriptor for Medicine Package Recognition

    Kenjiro Sugimoto, Koji Inoue, Yoshimitsu Kuroki, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E95D ( 5 ) 1264 - 1271  2012年05月  [査読有り]

     概要を見る

    This paper presents a color-based method for medicine package recognition, called a linear manifold color descriptor (LMCD). It describes a color distribution (a set of color pixels) of a color package image as a linear manifold (an affine subspace) in the color space, and recognizes an anonymous package by linear manifold matching. Mainly due to low dimensionality of color spaces, LMCD can provide more compact description and faster computation than description styles based on histogram and dominant-color. This paper also proposes distance-based dissimilarities for linear manifold matching. Specially designed for color distribution matching, the proposed dissimilarities are theoretically appropriate more than J-divergence and canonical angles. Experiments on medicine package recognition validates that LMCD outperforms competitors including MPEG-7 color descriptors in terms of description size, computational cost and recognition rate.

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • A generalized 3-D Hilbert scan using look-up tables

    Jian Zhang, Sei-ichiro Kamata

    JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION   23 ( 3 ) 418 - 425  2012年04月  [査読有り]

     概要を見る

    The Hilbert curve is a one-to-one mapping between multidimensional space and one-dimensional (1-D) space. Due to the advantage of preserving high correlation of multidimensional points, it receives much attention in many areas. Especially in image processing, Hilbert curve is studied actively as a scan technique (Hilbert scan). Currently there have been several Hilbert scan algorithms, but they usually have strict implementation conditions. For example, they use recursive functions to generate scans, which makes the algorithms complex and difficult to implement in real-time systems. Moreover the length of each side in a scanned region should be same and equal to the power of two, which limits the application of Hilbert scan greatly. In this paper, to remove the constraints and improve the Hilbert scan for a general application, an effective generalized three-dimensional (3-D) Hilbert scan algorithm is proposed. The proposed algorithm uses two simple look-up tables instead of recursive functions to generate a scan, which greatly reduces the computational complexity and saves storage memory. Furthermore, the experimental results show that the proposed generalized Hilbert scan can also take advantage of the high correlation between neighboring lattice points in an arbitrarily-sized cuboid region, and give competitive performance in comparison with some common scan techniques. (C) 2012 Elsevier Inc. All rights reserved.

    DOI

    Scopus

    8
    被引用数
    (Scopus)
  • Fast Hypercomplex Polar Fourier Analysis

    Zhuo Yang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E95D ( 4 ) 1166 - 1169  2012年04月  [査読有り]

     概要を見る

    Hypercomplex polar Fourier analysis treats a signal as a vector field and generalizes the conventional polar Fourier analysis. It can handle signals represented by hypercomplex numbers such as color images. Hypercomplex polar Fourier analysis is reversible that means it can reconstruct image. Its coefficient has rotation invariance property that can be used for feature extraction. However in order to increase the computation speed, fast algorithm is needed especially for image processing applications like realtime systems and limited resource platforms. This paper presents fast hypercomplex polar Fourier analysis based on symmetric properties and mathematical properties of trigonometric functions. Proposed fast hypercomplex polar Fourier analysis computes symmetric points simultaneously, which significantly reduce the computation time.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • INDIVIDUAL AUTHENTICATION THROUGH HAND POSTURE RECOGNITION USING MULTI-HILBERT SCANNING DISTANCE

    Jegoon Ryu, Sei-ichiro Kamata

    2012 PROCEEDINGS OF THE 20TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO)     1787 - 1790  2012年  [査読有り]

     概要を見る

    In this paper, we propose a novel Hand Posture Recognition (HPR) for biometrics. This study uses the three dimensional point clouds for robust hand posture recognition at the rotation and scale. Multi-Hilbert Scanning Distance (MHSD) are also introduced for mathematical approaches of shape matching. HPR framework is divided into five parts: detecting hand region, removing the wrist, aligning the hand pose, extracting feature descriptor, and matching. Based on the experimental results, this framework showed superior results for hand posture recognition rate.

  • A linear manifold color descriptor for medicine package recognition

    Kenjiro Sugimoto, Koji Inoue, Yoshimitsu Kuroki, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E95-D ( 5 ) 1264 - 1271  2012年

     概要を見る

    This paper presents a color-based method for medicine package recognition, called a linear manifold color descriptor (LMCD). It describes a color distribution (a set of color pixels) of a color package image as a linear manifold (an affine subspace) in the color space, and recognizes an anonymous package by linear manifold matching. Mainly due to low dimensionality of color spaces, LMCD can provide more compact description and faster computation than description styles based on histogram and dominant-color. This paper also proposes distance-based dissimilarities for linear manifold matching. Specially designed for color distribution matching, the proposed dissimilarities are theoretically appropriate more than J-divergence and canonical angles. Experiments on medicine package recognition validates that LMCD outperforms competitors including MPEG-7 color descriptors in terms of description size, computational cost and recognition rate. Copyright © 2012 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • Novel algorithm for polar and spherical fourier analysis on two and three dimensional images

    Zhuo Yang, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E95-D ( 5 ) 1248 - 1255  2012年

     概要を見る

    Polar and Spherical Fourier analysis can be used to extract rotation invariant features for image retrieval and pattern recognition tasks. They are demonstrated to show superiorities comparing with other methods on describing rotation invariant features of two and three dimensional images. Based on mathematical properties of trigonometric functions and associated Legendre polynomials, fast algorithms are proposed for multimedia applications like real time systems and large multimedia databases in order to increase the computation speed. The symmetric points are computed simultaneously. Inspired by relative prime number theory, systematic analysis are given in this paper. Novel algorithm is deduced that provide even faster speed. Proposed method are 9-15% faster than previous work. The experimental results on two and three dimensional images are given to illustrate the effectiveness of the proposed method. Multimedia signal processing applications that need real time polar and spherical Fourier analysis can be benefit from this work. Copyright © 2012 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

  • Image description with local patterns: An application to face recognition

    Wei Zhou, Alireza Ahrary, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E95-D ( 5 ) 1494 - 1505  2012年

     概要を見る

    In this paper, we propose a novel approach for presenting the local features of digital image using 1D Local Patterns by Multi-Scans (1DLPMS). We also consider the extentions and simplifications of the proposed approach into facial images analysis. The proposed approach consists of three steps. At the first step, the gray values of pixels in image are represented as a vector giving the local neighborhood intensity distrubutions of the pixels. Then, multi-scans are applied to capture different spatial information on the image with advantage of less computation than other traditional ways, such as Local Binary Patterns (LBP). The second step is encoding the local features based on different encoding rules using 1D local patterns. This transformation is expected to be less sensitive to illumination variations besides preserving the appearance of images embedded in the original gray scale. At the final step, Grouped 1D Local Patterns by Multi-Scans (G1DLPMS) is applied to make the proposed approach computationally simpler and easy to extend. Next, we further formulate boosted algorithm to extract the most discriminant local features. The evaluated results demonstrate that the proposed approach outperforms the conventional approaches in terms of accuracy in applications of face recognition, gender estimation and facial expression. Copyright © 2012 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

    10
    被引用数
    (Scopus)
  • SSM-HPC: Front view gait recognition using spherical space model with human point clouds

    Jegoon Ryu, Sei-Ichiro Kamata, Alireza Ahrary

    IEICE Transactions on Information and Systems   E95-D ( 7 ) 1969 - 1978  2012年

     概要を見る

    In this paper, we propose a novel gait recognition framework - Spherical Space Model with Human Point Clouds (SSM-HPC) to recognize front view of human gait. A new gait representation - Marching in Place (MIP) gait is also introduced which preserves the spatiotemporal characteristics of individual gait manner. In comparison with the previous studies on gait recognition which usually use human silhouette images from image sequences, this research applies three dimensional (3D) point clouds data of human body obtained from stereo camera. The proposed framework exhibits gait recognition rates superior to those of other gait recognition methods. © 2012 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • FAST IMAGE FILTERING BY DCT-BASED KERNEL DECOMPOSITION AND SEQUENTIAL SUM UPDATE

    Kenjiro Sugimoto, Sei-ichiro Kamata

    2012 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2012)     125 - 128  2012年  [査読有り]

     概要を見る

    This paper presents an approximate Gaussian filter which can run in one-pass with high accuracy based on spectrum sparsity. This method is a modification of the cosine integral image (CII), which decomposes a filter kernel into few cosine terms and convolves each cosine term with an input image in constant time per pixel by using integral images and look-up tables. However, they require much workspace and high access cost. The proposed method solves the problem with no decline in quality by sequentially updating sums instead of integral images and by improving look-up tables, which accomplishes a one-pass approximation with much less workspace. A specialization for tiny kernels are also discussed for faster calculation. Experiments on image filtering show that the proposed method can run nearly two times faster than CII and also than convolution even with small kernel.

  • Removal of transparent plastic film specular reflection based on multi-light sources

    Chengjie Wang, Sei-Ichiro Kamata

    2012 Symposium on Photonics and Optoelectronics, SOPO 2012    2012年

     概要を見る

    We present a novel method to remove the specular reflections on the surface of transparent plastic film. Our approach uses four light sources with strategic positions to get four images. Based on the information that both reflection and shadow move a lot from image to image, we reconstruct a high quality image free from reflection and shadow by using a image set which is consist of four images. © 2012 IEEE.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Registering 3D Objects Triangular Meshes using An Interest Point Detection on Barycentric Coordinates

    Tibyani Tibyani, Sei-ichiro Kamata

    2012 INTERNATIONAL CONFERENCE ON INFORMATICS, ELECTRONICS & VISION (ICIEV)     122 - 127  2012年  [査読有り]

     概要を見る

    In this paper, we put forward an interest point detection framework and combine with a spin image algorithm to register them. A framework is presented in this study. This method make use of the Harris Operator Extension method of interest point detection on 3D manifold triangular meshes in barycentric coordinate. Using this approach, we can extract the object correctly and effectively in noise situation. The unique advantage of this framework is its applicability to triangular meshes models. Experimental results on a different number of models are shown to demonstrate more accurate and effectively results for global registering 3D Objects triangular meshes for three pairs of corresponding interest point features.

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • A Fast and Accurate Interest Points Detection Algorithm on 3D Meshes using Extension of Harris Operator Combined with Hilbert Scanning Distance

    Tibyani Tibyani, Sei-ichiro Kamata

    2012 IEEE SYMPOSIUM ON INDUSTRIAL ELECTRONICS AND APPLICATIONS (ISIEA 2012)     367 - 371  2012年  [査読有り]

     概要を見る

    The main purpose of interest points detection (IPD) for 3D objects based on Harris operator is to find the fast computation of weighted average of the derivative data for different points. In this paper, we analyze the extension of Harris operator using Hausdorff distance (EHOHD) and propose the extension of Harris operator using Hilbert scanning distance (EHOHSD) as a new proposed method to IPD on 3D manifold triangular meshes data. Proposed EHOHSD method is 6-16 times faster than EHOHD method. The quality of this IPD with EHOHSD was measured using the repeatability criterion.

    DOI

    Scopus

  • An Efficient Window-Based Stereo Matching Algorithm using Foreground Disparity Concentration

    Xuejiao Bai, Sei-ichiro Kamata

    2012 12TH INTERNATIONAL CONFERENCE ON CONTROL, AUTOMATION, ROBOTICS & VISION (ICARCV)     1352 - 1357  2012年  [査読有り]

     概要を見る

    In this paper, we present an efficient window-based stereo matching algorithm that especially focuses on foreground objects. For decades, there are a lot of researches about the stereo matching algorithms. However, most of methods concentrate on the entire pixels, which are time consuming and meaningless in the real applications. To strength the accuracy of stereo correspondence in foreground objects, a simple locally support-weight method based on the selected prime key is proposed in our algorithm. Moreover, a background pre-detection method is also employed to get a primary background checking map, which is used to reduce the number of computed pixels in the disparity selection. After the refinement of both foreground disparity map and background checking map, our algorithm obtains accurate disparity results on the foreground and separate it with the background by the correspondence search simultaneously. The experimental results based on the Middlebury stereo datasets demonstrate that our method can achieve a better performance on foreground disparity computing than many other support-weight methods in terms of both accuracy and computational efficiency. In addition, our proposals can make foreground objects detection easier at the same time.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • FAST IMAGE FILTERING BY DCT-BASED KERNEL DECOMPOSITION AND SEQUENTIAL SUM UPDATE

    Kenjiro Sugimoto, Sei-ichiro Kamata

    2012 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP 2012)     125 - 128  2012年  [査読有り]

     概要を見る

    This paper presents an approximate Gaussian filter which can run in one-pass with high accuracy based on spectrum sparsity. This method is a modification of the cosine integral image (CII), which decomposes a filter kernel into few cosine terms and convolves each cosine term with an input image in constant time per pixel by using integral images and look-up tables. However, they require much workspace and high access cost. The proposed method solves the problem with no decline in quality by sequentially updating sums instead of integral images and by improving look-up tables, which accomplishes a one-pass approximation with much less workspace. A specialization for tiny kernels are also discussed for faster calculation. Experiments on image filtering show that the proposed method can run nearly two times faster than CII and also than convolution even with small kernel.

    DOI

    Scopus

    10
    被引用数
    (Scopus)
  • Unsupervised People Organization and Its Application on Individual Retrieval From Videos

    Pengyi Hao, Sei-ichiro Kamata

    2012 21ST INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR 2012)     2001 - 2004  2012年  [査読有り]

     概要を見る

    In this paper, a method named histogram intersection metric learning from scene tracks is proposed for automatic organizing people in videos. We make the following contributions: (i) learning histogram intersection distance instead of Mahalanobis distance for widely used face features; (ii) learning the metric from scene tracks without manually labeling any examples, which enables learning across large variations in pose, expression, occlusion and illumination with small number of face pairs and can distinguish different people powerfully. We firstly test face identification, track clustering, and people organization on a long film, then individual retrieval based on people organization from a large video dataset is evaluated, demonstrating significantly increased search quality with respect to previous approaches on this area.

  • Face representation and recognition with local curvelet patterns

    Wei Zhou, Alireza Ahrary, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E95-D ( 12 ) 3078 - 3087  2012年

     概要を見る

    In this paper, we propose Local Curvelet Binary Patterns (LCBP) and Learned Local Curvelet Patterns (LLCP) for presenting the local features of facial images. The proposed methods are based on Curvelet transform which can overcome the weakness of traditional Gabor wavelets in higher dimensions, and better capture the curve singularities and hyperplane singularities of facial images. LCBP can be regarded as a combination of Curvelet features and LBP operator while LLCP designs several learned codebooks from patch sets, which are constructed by sampling patches from Curvelet filtered facial images. Each facial image can be encoded into multiple pattern maps and block-based histograms of these patterns are concatenated into an histogram sequence to be used as a face descriptor. During the face representation phase, one input patch is encoded by one pattern in LCBP while multi-patterns in LLCP. Finally, an effective classifier called Weighted Histogram Spatially constrained Earth Mover's Distance (WHSEMD) which utilizes the discriminative powers of different facial parts, the different patterns and the spatial information of face is proposed. Performance assessment in face recognition and gender estimation under different challenges shows that the proposed approaches are superior than traditional ones. Copyright © 2012 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

  • Fast Polar Cosine Transform for image description

    Yang, Zhuo, Kamata, Sei Ichiro

    Proceedings of the 12th IAPR Conference on Machine Vision Applications, MVA 2011     320 - 323  2011年12月

     概要を見る

    Polar Cosine Transform (PCT) is one of the Polar Harmonic Transforms that those kernels are basic waves and harmonic in nature. They are proposed to represent invariant patterns for two dimensional image description and are demonstrated to show superiorities comparing with other methods on extracting rotation invariant patterns for images. However in order to increase the computation speed, fast algorithm for PCT is proposed for real world applications like limited computing environments, large image databases and realtime systems. Based on our previous work, this paper novelly employs relative prime number theory to develop Fast Polar Cosine Transform (FPCT). The proposed FPCT is averagely over 11 ∼ 12.5 times faster than PCT that significantly boost computation process. The experimental results are given to illustrate the effectiveness of the proposed method.

  • Fast color matching using weighted subspace on medicine package recognition

    Sugimoto, Kenjiro, Kamata, Sei Ichiro

    Proceedings of the 12th IAPR Conference on Machine Vision Applications, MVA 2011     287 - 290  2011年12月

     概要を見る

    This paper presents a color matching technique using weighted subspace on medicine package recognition. The proposed method is more compact and lowercomplex than scalable color descriptor and dominant color descriptor, which are employed by MPEG-7. Our method is based on subspace matching: A color object is treated as a subspace derived from its color distribution. Unlike mutual subspace method, it is specially designed for color matching. Specifically, weighted subspace and a distance-based dissimilarity are employed instead of normalized subspace and similarity based on canonical angles of MSM. Experiments show that the proposed method outperforms the conventional methods in terms of description size, building/matching speed, and recognition rate.

    CiNii

  • Manifold learning based on multi-feature for road-sign recognition

    Zhang, Qieshi, Zhang, Qieshi, Kamata, Sei Ichiro

    Proceedings of the SICE Annual Conference     1143 - 1146  2011年11月

     概要を見る

    In this paper, a multi-feature selection and application based manifold learning metric method is proposed for Road-Sign Recognition (RSR). Firstly, the manifold metric between manifold from subspace is discussed in detail. After that, the multi-feature analyzing, selection, classification and application are introduced for rough recognition and create the manifold. Then the proposed method is used to evaluate the distance between the manifolds. Finally, the RSR results suggest that the proposed method is robust than other methods. © 2011 SICE.

  • Hypercomplex Polar Fourier Analysis for Image Representation

    Zhuo Yang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E94D ( 8 ) 1663 - 1670  2011年08月  [査読有り]

     概要を見る

    Fourier transform is a significant tool in image processing and pattern recognition. By introducing a hypercomplex number, hypercomplex Fourier transform treats a signal as a vector field and generalizes the conventional Fourier transform. Inspired from that, hypercomplex polar Fourier analysis that extends conventional polar Fourier analysis is proposed in this paper. The proposed method can handle signals represented by hypercomplex numbers as color images. The hypercomplex polar Fourier analysis is reversible that means it can be used to reconstruct image. The hypercomplex polar Fourier descriptor has rotation invariance property that can be used for feature extraction. Due to the noncommutative property of quaternion multiplication, both left-side and right-side hypercomplex polar Fourier analysis are discussed and their relationships are also established in this paper. The experimental results on image reconstruction, rotation invariance, color plate test and image retrieval are given to illustrate the usefulness of the proposed method as an image analysis tool.

    DOI

    Scopus

    4
    被引用数
    (Scopus)
  • Hilbert Scan Based Bag-of-Features for Image Retrieval

    Pengyi Hao, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E94D ( 6 ) 1260 - 1268  2011年06月  [査読有り]

     概要を見る

    Generally, two problems of bag-of-features in image retrieval are still considered unsolved: one is that spatial information about descriptors is not employed well, which affects the accuracy of retrieval; the other is that the trade-off between vocabulary size and good precision, which decides the storage and retrieval performance. In this paper, we propose a novel approach called Hilbert scan based bag-of-features (HS-BoF) for image retrieval. Firstly, Hilbert scan based tree representation (HSBT) is studied, which is built based on the local descriptors while spatial relationships are added into the nodes by a novel grouping rule, resulting of a tree structure for each image. Further, we give two ways of codebook production based on HSBT: multi-layer codebook and multi-size codebook. Owing to the properties of Hilbert scanning and the merits of our grouping method, sub-regions of the tree are not only flexible to the distribution of local patches but also have hierarchical relations. Extensive experiments on caltech-256, 13-scene and 1 million Image Net images show that HS-BoF obtains higher accuracy with less memory usage.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Color Matching Using Weighted Subspace

    Kenjiro SUGIMOTO, Sei-ichiro KAMATA

    Proc. of 12th. IAPR Conference on Machine Vision Applications (MVA2011)     287 - 290  2011年06月

  • Fast Polar Cosine Transform for Image Description

    Zhuo YANG, Sei-ichiro KAMATA

    Proc. of 12th. IAPR Conference on Machine Vision Applications (MVA2011)     320 - 323  2011年06月

  • B-splineに基づくContourletスパース表現を用いた画像ノイズ除去

    呉 仁傑, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2011   23 - 23  2011年

     概要を見る

    Abstract 画像強調,ノイズ除去において,画像空間を周波数領域に変換し,分解された高周波数成分を解析する手法がある.従来の変換において,Fourier変換やWavelet変換では画像におけるエッジの幾何特徴をとらえるのに限界があることが知られている.本稿では,従来より良い近似性能を得るために,B-spline曲線に基づくContourlet変換スパース表現手法を提案し,ランダムノイズに対する除去法を検討した.Keywords Contourlet変換,B-spline曲線、スパース表現

    DOI CiNii

  • HYPERCOMPLEX POLAR FOURIER ANALYSIS FOR COLOR IMAGE

    Zhuo Yang, Sei-ichiro Kamata

    2011 18TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     2161 - 2164  2011年  [査読有り]

     概要を見る

    Fourier transform is a significant tool in image processing and pattern recognition. By introducing hyper complex number, hyper complex Fourier transform [1] treats signal as vector field and generalizes conventional Fourier transform. Inspired from that, hypercomplex polar Fourier analysis is proposed in this paper. This work extends conventional polar Fourier analysis [5]. The proposed method can handle hypercomplex number represented signals like color image. The hypercomplex polar Fourier analysis is reversible that means it can be used to reconstruct image. The hypercomplex polar Fourier descriptor has rotation invariance property that can be used for feature extraction. Due to the noncommutative property of quaternion multiplication, both left-side and right-side hypercomplex polar Fourier analysis are discussed and their relationships are also established in this paper. The experimental results on image reconstruction, rotation invariance and color plate test are given to illustrate the usefulness of the proposed method as an image analysis tool.

  • FRONT VIEW GAIT RECOGNITION USING SPHERICAL SPACE MODEL WITH HUMAN POINT CLOUDS

    Jegoon Ryu, Sei-ichiro Kamata

    2011 18TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     3270 - 3273  2011年  [査読有り]

     概要を見る

    In this paper, we propose a novel gait recognition framework which is Spherical Space Model with Human Point Clouds (SSM-HPC). A new gait representation is also introduced, which is called Marching in Place (MIP) gait and preserves the spatiotemporal characteristics of individual gait manner. Various researches for gait recognition have used human silhouette images from moving picture. This research uses Three Dimensional (3D) point clouds data of human body obtained from stereo camera, which has the scale-invariant property. The framework is applied for frontal view gait recognition. This framework showed superior results for gait recognition rate than other gait recognition methods.

  • COLOR DISTRIBUTION MATCHING USING A WEIGHTED SUBSPACE DESCRIPTOR

    Kenjiro Sugimoto, Sei-ichiro Kamata

    2011 18TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     1697 - 1700  2011年  [査読有り]

     概要を見る

    This paper presents a low-level color descriptor which describes the color distribution of a color image as a weighted subspace in the color space, namely eigenvectors and eigen-values of the distribution. Thanks to low-dimensionality of color space, the proposed descriptor can provide compact description and fast computation. Furthermore, specialized for color distribution matching, it is more efficient than mutual subspace method (MSM). Experiments on medicine package recognition validate that the proposed descriptor outperforms MSM and MPEG-7 low-level color descriptors in terms of description size, computational cost and recognition rate.

  • SINGLE IMAGE BASED HAZE REMOVAL METHOD

    Qieshi Zhang, Sei-Ichiro Kamata

    FOURTH INTERNATIONAL CONFERENCE ON COMPUTER AND ELECTRICAL ENGINEERING (ICCEE 2011)     365 - 369  2011年  [査読有り]

     概要を見る

    In this paper, we propose a novel method to estimate the haze-free image based on the improved optical model. In this work, we calculate the objective transmission and distance transmission to estimate the haze-free image by the segmented hazy image. Firstly, the color clustering method is used to segment the image into several regions by color similarity for getting the objective transmission. Then, the graph-based segmentation is used to calculate the depth information for getting the distance transmission. Next, the atmosphere light is estimated according to the distance transmission Finally, the improved optical model is used to estimate the haze-free image. The experimental results show that our method is more effective and able to get better results than other single image based methods.

  • Anisotropic diffusion with edge projection

    Gang Qiao, Wei Zhang, Sei-Ichiro Kamata

    Proceedings - 4th International Congress on Image and Signal Processing, CISP 2011   2   665 - 667  2011年

     概要を見る

    In image processing, anisotropic diffusion provides a forward method to remove noise while preserving edges accurate and sharp. However, due to the inappropriate edge estimation by gradient, some isolated noise points still exist and edge location is inaccurate. In this representation, isolated noise points and edges are distinguished by the significant difference of their "lengths", which are computed by orthogonally projecting their pixels to the corresponding normalized gradient directions and recording the number of the same projections. Combining gradient and "length" to estimate edges, isolated noise points are further suppressed while edges are re-located and enhanced. © 2011 IEEE.

    DOI

    Scopus

  • Fast Hypercomplex Polar Fourier Analysis for Image Processing

    Zhuo Yang, Sei-ichiro Kamata

    ADVANCES IN IMAGE AND VIDEO TECHNOLOGY, PT II   7088 ( PART 2 ) 141 - 148  2011年  [査読有り]

     概要を見る

    Hypercomplex polar Fourier analysis treats a signal as a vector field and generalizes the conventional polar Fourier analysis. It can handle signals represented by hypercomplex numbers such as color images. It is reversible that can reconstruct image. Its coefficient has rotation invariance property that can be used for feature extraction. With these properties, it can be used for image processing applications like image representation and image understanding. However in order to increase the computation speed, fast algorithm is needed especially for image processing applications like realtime systems and limited resource platforms. This paper presents fast hypercomplex polar Fourier analysis that based on symmetric properties and mathematical properties of trigonometric functions. Proposed fast hypercomplex polar Fourier analysis computes symmetric eight points simultaneously that significantly reduce the computation time.

    DOI

    Scopus

  • Multi balanced trees for face retrieval from image database

    Pengyi Hao, Sei-Ichiro Kamata

    2011 IEEE International Conference on Signal and Image Processing Applications, ICSIPA 2011     484 - 489  2011年

     概要を見る

    We are interested here in retrieving images containing a specific person in image database. Due to large variations in illumination conditions, hairstyles, facial expressions, etc. and the factors like occlusion, sunglasses, profile, etc., robust face matching has been a challenging problem. On the other hand, the speed of search is also a considerable issue, especially for the dataset with millions of face images. Inspired by face tracks in video retrieval which take advantages from the abundance of frames to get multiple exemplars, we present an approach named multi balanced trees for face retrieval from image dataset in this paper. Face images in the dataset are efficiently organized by the trees produced for persons. Multi sampling on the facial components employs the rich local information, which can help to differentiate different persons. Given a query face, a sorted face set with similarities is obtained by inserting the query into a tree. It is easy and fast to get the search results in respect that it avoids calculating the distances between query and elements in the cluster. In addition, a rectification strategy is given in the query process to rectify the error occurred in the generation of trees, resulting in a significant improvement of retrieval quality. Experimental results show the better face grouping ability in comparison with traditional methods. The speed of searching is improved as well. © 2011 IEEE.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • An improved method for illumination invariant face recognition based on adaptive rescaling DCT coefficient in logarithm domain

    Chao Yu, Xiaoqun Zhao, Sei-Ichiro Kamata

    Lecture Notes in Electrical Engineering   121   297 - 304  2011年

     概要を見る

    This paper presents an improved method for robust face recognition using illumination normalization based on Discrete Cosine Transform (DCT) in logarithm domain. Two novel coefficients are designed to identify the lighting condition (LC), based on which the low-frequency DCT coefficients are adaptively rescaled except the first one (DC). As a result variations under different illumination conditions are minimized meanwhile original information contained in low-frequency is comparatively well preserved. Results of experiments on Yale B database and Extended Yale B database show that proposed method has better performance under variational input illumination conditions. The proposed method is fast in computation and could be easily implemented into real time face recognition systems. © 2011 Springer-Verlag.

    DOI

    Scopus

  • Fast Hypercomplex Polar Fourier Analysis for Image Processing

    Zhuo Yang, Sei-ichiro Kamata

    ADVANCES IN IMAGE AND VIDEO TECHNOLOGY, PT II   7088   141 - 148  2011年  [査読有り]

     概要を見る

    Hypercomplex polar Fourier analysis treats a signal as a vector field and generalizes the conventional polar Fourier analysis. It can handle signals represented by hypercomplex numbers such as color images. It is reversible that can reconstruct image. Its coefficient has rotation invariance property that can be used for feature extraction. With these properties, it can be used for image processing applications like image representation and image understanding. However in order to increase the computation speed, fast algorithm is needed especially for image processing applications like realtime systems and limited resource platforms. This paper presents fast hypercomplex polar Fourier analysis that based on symmetric properties and mathematical properties of trigonometric functions. Proposed fast hypercomplex polar Fourier analysis computes symmetric eight points simultaneously that significantly reduce the computation time.

  • Hypercomplex polar Fourier analysis for color image

    Zhuo Yang, Sei-Ichiro Kamata

    Proceedings - International Conference on Image Processing, ICIP     2117 - 2120  2011年

     概要を見る

    Fourier transform is a significant tool in image processing and pattern recognition. By introducing hypercomplex number, hypercomplex Fourier transform [1] treats signal as vector field and generalizes conventional Fourier transform. Inspired from that, hypercomplex polar Fourier analysis is proposed in this paper. This work extends conventional polar Fourier analysis [5]. The proposed method can handle hypercomplex number represented signals like color image. The hypercom-plex polar Fourier analysis is reversible that means it can be used to reconstruct image. The hypercomplex polar Fourier descriptor has rotation invariance property that can be used for feature extraction. Due to the noncommutative property of quaternion multiplication, both left-side and right-side hypercomplex polar Fourier analysis are discussed and their relationships are also established in this paper. The experimental results on image reconstruction, rotation invariance and color plate test are given to illustrate the usefulness of the proposed method as an image analysis tool. © 2011 IEEE.

    DOI

    Scopus

    11
    被引用数
    (Scopus)
  • Multi balanced trees for face retrieval from image database

    Pengyi Hao, Sei-Ichiro Kamata

    2011 IEEE International Conference on Signal and Image Processing Applications, ICSIPA 2011     484 - 489  2011年

     概要を見る

    We are interested here in retrieving images containing a specific person in image database. Due to large variations in illumination conditions, hairstyles, facial expressions, etc. and the factors like occlusion, sunglasses, profile, etc., robust face matching has been a challenging problem. On the other hand, the speed of search is also a considerable issue, especially for the dataset with millions of face images. Inspired by face tracks in video retrieval which take advantages from the abundance of frames to get multiple exemplars, we present an approach named multi balanced trees for face retrieval from image dataset in this paper. Face images in the dataset are efficiently organized by the trees produced for persons. Multi sampling on the facial components employs the rich local information, which can help to differentiate different persons. Given a query face, a sorted face set with similarities is obtained by inserting the query into a tree. It is easy and fast to get the search results in respect that it avoids calculating the distances between query and elements in the cluster. In addition, a rectification strategy is given in the query process to rectify the error occurred in the generation of trees, resulting in a significant improvement of retrieval quality. Experimental results show the better face grouping ability in comparison with traditional methods. The speed of searching is improved as well. © 2011 IEEE.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • A common key encryption algorithm using N-dimensional Hilbert curves

    Sei-Ichiro Kamata

    Proceedings of the 2011 7th International Conference on Information Assurance and Security, IAS 2011     275 - 279  2011年

     概要を見る

    There are a lot of previous works on common key encryptions such as DES, AES, etc, In this paper, a new common key encryption algorithm is proposed using Hilbert curves which are a one-to-one mapping between N-dimensional (N-D) spaces and 1-D space (a line). This is based on a property having a sharp rise in the number of Hilbert curve patterns in N-D spaces. In the case of N = 2, there are only four patterns, while if N is 5, the number of the patterns is more than 1 billions. Operations of addition and multiplication are denned on a curve, based on a mapping of a point in N-D spaces to a point on a line. In order to realize a cryptosystem, the algorithm utilizes Hilbert ordered point addresses, which is expressed as the coordinates of the points in N-dimensional space. © 2011 IEEE.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • FRONT VIEW GAIT RECOGNITION USING SPHERICAL SPACE MODEL WITH HUMAN POINT CLOUDS

    Jegoon Ryu, Sei-ichiro Kamata

    2011 18TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     3209 - 3212  2011年  [査読有り]

     概要を見る

    In this paper, we propose a novel gait recognition framework which is Spherical Space Model with Human Point Clouds (SSM-HPC). A new gait representation is also introduced, which is called Marching in Place (MIP) gait and preserves the spatiotemporal characteristics of individual gait manner. Various researches for gait recognition have used human silhouette images from moving picture. This research uses Three Dimensional (3D) point clouds data of human body obtained from stereo camera, which has the scale-invariant property. The framework is applied for frontal view gait recognition. This framework showed superior results for gait recognition rate than other gait recognition methods.

    DOI

    Scopus

    17
    被引用数
    (Scopus)
  • COLOR DISTRIBUTION MATCHING USING A WEIGHTED SUBSPACE DESCRIPTOR

    Kenjiro Sugimoto, Sei-ichiro Kamata

    2011 18TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)     1697 - 1700  2011年  [査読有り]

     概要を見る

    This paper presents a low-level color descriptor which describes the color distribution of a color image as a weighted subspace in the color space, namely eigenvectors and eigen-values of the distribution. Thanks to low-dimensionality of color space, the proposed descriptor can provide compact description and fast computation. Furthermore, specialized for color distribution matching, it is more efficient than mutual subspace method (MSM). Experiments on medicine package recognition validate that the proposed descriptor outperforms MSM and MPEG-7 low-level color descriptors in terms of description size, computational cost and recognition rate.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • An improved method for illumination invariant face recognition based on adaptive rescaling DCT coefficient in logarithm domain

    Chao Yu, Xiaoqun Zhao, Sei-Ichiro Kamata

    Lecture Notes in Electrical Engineering   121   297 - 304  2011年

     概要を見る

    This paper presents an improved method for robust face recognition using illumination normalization based on Discrete Cosine Transform (DCT) in logarithm domain. Two novel coefficients are designed to identify the lighting condition (LC), based on which the low-frequency DCT coefficients are adaptively rescaled except the first one (DC). As a result variations under different illumination conditions are minimized meanwhile original information contained in low-frequency is comparatively well preserved. Results of experiments on Yale B database and Extended Yale B database show that proposed method has better performance under variational input illumination conditions. The proposed method is fast in computation and could be easily implemented into real time face recognition systems. © 2011 Springer-Verlag.

    DOI

    Scopus

  • A Fast Homology Search Algorithm Using Dynamic Seeding

    Haijiang TANG, Sei-ichiro KAMATA, Toshimasa YAMAZAKI

    Proc. of The 21th International Conference on Genome Informatics    2010年12月

  • A Color Distribution Descriptor for Medicine Package Recognition

    Kenjiro Sugimoto, Koji Inoue, Kuroki Yoshimitsu, Sei-ichiro Kamata

    Proc. of 2nd China-Japan-Korea Joint Workshop of Pattern Recognition    2010年11月

  • A Study on Fast Random Access Decompression Using Start-step-stop Coding and Rank/Select Dictionary

    Kenjiro SUGIMOTO, Sei-ichiro KAMATA

    Proc. of 1st Int. Conf. on Advanced Computing and Communications (ACC-2010)     132 - 135  2010年09月

  • Interscale Stein's Unbiased Risk Estimate and Intrascale Feature Patches Distance Constraint for Image Denoising

    Qieshi Zhang, Sei-ichiro Kamata, Alireza Ahrary

    IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES   E93A ( 8 ) 1434 - 1441  2010年08月  [査読有り]

     概要を見る

    The influence of noise is an important problem on image acquisition and transmission stages. The traditional image denoising approaches only analyzing the pixels of local region with a moving window, which calculated by neighbor pixels to denoise. Recently, this research has been focused on the transform domain and feature space. Compare with the traditional approaches, the global multi-scale analyzing and unchangeable noise distribution is the advantage. Apparently, the estimation based methods can be used in transform domain and get better effect. This paper proposed a new approach to image denoising in orthonormal wavelet domain. In this paper, we adopt Stein's unbiased risk estimate (SURE) based method to denoise the low-frequency bands and the feature patches distance constraint (FPDC) method also be proposed to estimate the noise free bands in Wavelet domain. The key point is that how to divide the lower frequency sub-bands and the higher frequency sub-bands, and do interscale SURE and intrascale FPDC, respectively. We compared our denoising method with some well-known and new denoising algorithms, the experimental results show that the proposed method can give better performance and keep more detail information in most objective and subjective criteria than other methods.

    DOI

    Scopus

  • Stein's unbiased risk estimate (SURE) and distance constraint combined image denoising in Wavelet domain

    Zhang, Qieshi, Kamata, Sei Ichiro

    Proceedings of the 7th IASTED International Conference on Signal Processing, Pattern Recognition and Applications, SPPRA 2010     196 - 201  2010年07月

     概要を見る

    Image denoising is a lively research field now. For solving this problem, non-linear filters based methods are the classical approach. These methods are based on local analysis of pixels with a moving window in spatial domain, but also have some shortcoming. Recently, because of the properties of Wavelet transform, this research has been focused on the wavelet domain. Compared to the classical nonlinear filters, the global multi-scale analysis characteristic of Wavelet is better for image denoising. So this paper proposed a new approach to use orthonormal Wavelet transform and distance constraint to solve this. Here, by minimizing the Stein's unbiased risk estimate (SURE) method to calculate the low frequency sub-band images for estimating. And convert the high frequency sub-band images to feature space, then use distance constraint to denoise by trained samples set. The experimental results show that the proposed method is efficiency and keep the detail ideally.

  • Fast Polar Harmonic Transforms

    Zhou YANG, Alireza AHRARY, Sei-ichiro KAMATA

    The Journal of the IIEEJ   39 ( 4 ) 399 - 408  2010年07月

  • Fast Polar and Spherical Fourier Descriptors for Feature Extraction

    Zhuo Yang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E93D ( 7 ) 1708 - 1715  2010年07月

     概要を見る

    Polar Fourier Descriptor(PFD) and Spherical Fourier Descriptor(SFD) are rotation invariant feature descriptors for two dimensional(2D) and three dimensional(3D) image retrieval and pattern recognition tasks. They are demonstrated to show superiorities compared with other methods on describing rotation invariant features of 2D and 3D images. However in order to increase the computation speed, fast computation method is needed especially for machine vision applications like realtime systems, limited computing environments and large image databases. This paper presents fast computation method for PFD and SFD that are deduced based on mathematical properties of trigonometric functions and associated Legendre polynomials. Proposed fast PFD and SFD are 8 and 16 times faster than direct calculation that significantly boost computation process. Furthermore, the proposed methods are also compact for memory requirements for storing PFD and SFD basis in lookup tables. The experimental results on both synthetic and real data are given to illustrate the efficiency of the proposed method.

    DOI

    Scopus

    4
    被引用数
    (Scopus)
  • An adaptive tone mapping algorithm for high dynamic range images

    Jian Zhang, Sei-Ichiro Kamata

    Kyokai Joho Imeji Zasshi/Journal of the Institute of Image Information and Television Engineers   64 ( 6 ) 850 - 860  2010年06月

     概要を見る

    A common task of tone mapping algorithms is to reproduce high dynamic range images (HDR) on low dynamic range (LDR) display devices such as printers and monitors. We present a new tone mapping algorithm for the display of HDR images that was inspired by the adaptive process of the human visual system. The proposed algorithm is based on center/surround Retinex processing. Our method has two novel aspects. The input luminance image is first compressed by a global tone mapping curve. The curvature of the compression curve is adapted locally based on the pseudo-Hilbert scan technique, so it can provide a better overall impression before the subsequent local processing. Second, the local details are enhanced according to a non-linear adaptive spatial filter (Gaussian filter), whose shape (filter variance) is adapted to the high-contrast edges of the image. The proposed method takes advantage of the properties of both global and local processing while overcoming their respective disadvantages. Therefore, the algorithm can preserve visibility and contrast impression of high dynamic range scenes in standard display devices. We tested the proposed method on a variety of HDR images and also compared it to previous research. The results indicated that our method was effective for displaying images with high visual quality.

    DOI

    Scopus

  • 高ダイナミックレンジ画像トーンマッピング

    張 兼, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2010 ( 0 ) 312 - 313  2010年

     概要を見る

    A common task of tone mapping is to reproduce high dynamic range images (HDR) on low dynamic range (LDR) display devices such as printers and monitors. In this paper, a new tone mapping algorithm is proposed. Compared to the previous algorithms, our approach uses an adaptive surround instead of the traditional pre-defined circular. So the shape of a surround can be changed according to the high-contrast edges, which can effectively avoid halo artifacts but preserve visibility of local details. The experimental results show that this algorithm is effective and easy to use.

    CiNii

  • ポイントクラウドと球面空間モデルを利用した前方視点からの歩容認識

    柳 済群, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2010 ( 0 ) 311 - 312  2010年

     概要を見る

    In this paper, we propose a novel gait recognition framework which is spherical space model using human point clouds (SSM-HPC). The framework is applied for frontal view gait recognition.Various researches for gait recognition have been used human silhouette images from moving picture. This research used three dimensional point cloud data from stereo camera. This framework can get good result from gait recognition rate than silhouette image.

    CiNii

  • Pixel Color Feature Enhancement for Road Signs Detection

    Qieshi Zhang, Sei-ichiro Kamata

    SECOND INTERNATIONAL CONFERENCE ON DIGITAL IMAGE PROCESSING   7546  2010年  [査読有り]

     概要を見る

    Road signs play an important role in our daily life which used to guide drivers to notice variety of road conditions and cautions. They provide important visual information that can help drivers operating their vehicles in a manner for enhancing traffic safety. The occurrence of some accidents can be reduced by using automatic road signs recognition system which can alert the drivers. This research attempts to develop a warning system to alert the drivers to notice the important road signs early enough to refrain road accidents from happening. For solving this, a non-linear weighted color enhancement method by pixels is presented. Due to the advantage of proposed method, different road signs can be detected from videos effectively. With suitably coefficients and operations, the experimental results have proved that the proposed method is robust, accurate and powerful in road signs detection.

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • On-line Signature Matching Based on Hilbert Scanning Patterns

    Alireza Ahrary, Jian Zhang, Sei-ichiro Kamata

    Journal of the Institute of Image Electronics Engineers of Japan   39 ( 2 ) 175 - 184  2010年

     概要を見る

    Signature verification is a challenging task, because only a small set of genuine samples can be acquired and usually no forgeries are available in real application. In this paper, we propose a novel approach based on Hilbert scanning patterns and Gaussian mixture models for automatic on-line signature verification. Our system is composed of a similarity measure based on Hilbert scanning patterns and a simplified Gaussian mixture model for decision-level evaluation. To be practical, we introduce specific simplification ways for constructing a model and its training method. The system is compared to other state-of-the-art systems based on the results of the First International Signature Verification Competition (SVC 2004). Experiments are conducted to verify the effectiveness of our system. © 2010, The Institute of Image Electronics Engineers of Japan. All rights reserved.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Fast polar and spherical fourier descriptors for feature extraction

    Zhuo Yang, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E93-D ( 7 ) 1708 - 1715  2010年

     概要を見る

    Polar Fourier Descriptor(PFD) and Spherical Fourier Descriptor( SFD) are rotation invariant feature descriptors for two dimensional( 2D) and three dimensional(3D) image retrieval and pattern recognition tasks. They are demonstrated to show superiorities compared with other methods on describing rotation invariant features of 2D and 3D images. However in order to increase the computation speed, fast computation method is needed especially for machine vision applications like realtime systems, limited computing environments and large image databases. This paper presents fast computation method for PFD and SFD that are deduced based on mathematical properties of trigonometric functions and associated Legendre polynomials. Proposed fast PFD and SFD are 8 and 16 times faster than direct calculation that significantly boost computation process. Furthermore, the proposed methods are also compact for memory requirements for storing PFD and SFD basis in lookup tables. The experimental results on both synthetic and real data are given to illustrate the efficiency of the proposed method. Copyright © 2010 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

    4
    被引用数
    (Scopus)
  • Fast polar and spherical fourier descriptors for feature extraction

    Zhuo Yang, Sei-Ichiro Kamata

    Proceedings - International Conference on Pattern Recognition     975 - 978  2010年

     概要を見る

    Polar Fourier Descriptor(PFD) and Spherical Fourier Descriptor(SFD) are rotation invariant feature descriptors for two dimensional(2D) and three dimensional(3D) image retrieval and pattern recognition tasks. They are demonstrated to show superiorities compared with other methods on describing rotation invariant features of 2D and 3D images. However in order to increase the computation speed, fast computation method is needed especially for applications like realtime systems and large image databases. This paper presents fast computation method for PFD and SFD that based on mathematical properties of trigonometric functions and associated Legendre polynomials. Proposed fast PFD and SFD are 8 and 16 times faster than traditional ones that significantly boost computation process. © 2010 IEEE.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • 3D OBJECT MATCHING BASED ON SPHERICAL HILBERT SCANNING

    Can Tong, Sei-ichiro Kamata

    2010 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING     2941 - 2944  2010年  [査読有り]

     概要を見る

    This paper describes a novel method to match objects in cluttered scenes. This method makes use of Hilbert scanning of feature points in Hough space. We use a 3D Hough transform to obtain a spectrum on which 3D features are concentrated on the sphere. Then, based on the obtained Hough Spectrum, we apply Hilbert scanning on the sphere to match the objects. Using this approach, we can match the object correctly and robustly in both overlapping and noise situation. The characteristic of this method is that it is a global matching method without an estimate of the rotation first and suffering from computational complexity brought by voting/correlation procedure. The experiment results show that the method is more effective compared to existing methods in both matching rate and robustness.

  • IMAGE DESCRIPTION WITH 1D LOCAL PATTERNS BY MULTI-SCANS: AN APPLICATION TO FACE RECOGNITION

    Wei Zhou, Alireza Ahrary, Sei-ichiro Kamata

    2010 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING     4553 - 4556  2010年  [査読有り]

     概要を見る

    In this work, a novel approach which called 1D local patterns by multi-scans(1DLPMS) for presenting the local features is proposed and its simplifications and extensions to facial image analysis are also considered. First, multi-scans are applied to capture different spatial information on the image with less computation than some traditional ways, such as Local Binary Patterns(LBP). Then, some 1D local patterns are given to encode the local features based on different coding rules. To make the proposed approach computationally simpler and easy to extend, grouped 1D local patterns by multiscans(G1DLPMS) is studied, which divides 1DLPMS into several groups and uses the co-occurrences of these groups. Performance assessment in face recognition under different challenges shows that the proposed approach is superior than traditional ones.

  • Document Layout Analysis and Reading Order Determination for a Reading Robot

    Yucun Pan, Qunfei Zhao, Seiichiro Kamata

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     1607 - 1612  2010年  [査読有り]

     概要を見る

    In this paper an efficient approach of document layout analysis and reading order determination is proposed for a reading robot. Firstly the input document images are preprocessed to remove noises, connect lines and domains, and to reduce the computation time. Secondly a bottom-up, parameter-independent, two-step layout analysis algorithm based on morphology is used, which outlines the geometry of the maximum homogeneous regions and classifies them into texts, tables, and pictures. Finally the reading order is determined, by a top-down recursive hierarchy algorithm derived from XY-cut, using a set of rules depending on layout information. Important parameters are acquired using statistic information of the given images to adapt to different types of documents. The proposed algorithm is applied to a large number of document images and the experimental results show that it makes the reading robot be able to read paper documents of different languages, even with complex layout structure.

  • 3D Reconstruction from a Single Image for a Chinese Talking Face

    Ning Liu, Ning Fang, Seiichiro Kamata

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     1613 - 1616  2010年  [査読有り]

     概要を見る

    This paper proposes an automatic 3D reconstruction approach for a Chinese talking face by a generic model and a single image. Firstly, an improved color-based ASM method is used to detect the face area and get the 2D face feature points automatically from the given image, which is not restricted to full frontal one. Then, color information is used to correct the location of face feature points. Finally, after text mapping, a particular and realistic 3D face model is deformed from a generic model. Using ASM face feature points extraction and correction based on skin color model, the problem of side face information missing is successfully resolved. Depending on only one image and one generic model, the computing cost of memory and time is largely reduced. The 3D face reconstructed can be easily deformed to form different expressions and mouth shapes. Experiments show that this approach is fast and efficient and has an output of a lifelike Chinese talking face.

  • Face Detection in Color Images Based on Skin Color Models

    Li Zou, Sei-ichiro Kamata

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     681 - 686  2010年  [査読有り]

     概要を見る

    Face finding is a very important initial step towards building up a fully automated face recognition system. Face detection by detecting skin like colors can achieve a high detection rate. In this paper, we presented a novel algorithm for face detection in color images with complex backgrounds. First a parallel structure for skin color detection is proposed to improve the accuracy of detections. The concept of the probability image has been introduced to utilize the color information in the traditional face detection methods specific for gray-scale images. After that, a classifier obtained from Adaboost training is applied to the result of skin detection to reduce the false positives. An experiment has been implemented to verify the improvement of this proposed research. And the proposed approach achieved a better result in this experiment.

  • A Novel Face Representation Toward Pose Invariant Face Recognition

    Liang Yu, Sei-ichiro Kamata, Yong Fang

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     179 - 183  2010年  [査読有り]

     概要を見る

    Though elastic bunch graph matching (EBGM) has a good performance on face recognition in the distortion of facial expression, it is still not robust enough to in-depth rotation. To solve this problem, a novel face representation approach based on the space-filling tree is proposed in this paper. This kind of representation shows a better performance than Elastic bunch graph matching (EBGM) in in-depth rotation of pose especially when there are only frontal images in the training set. With the proposed face representation approach, the face recognition system is built. Experimental results on the FERET standard database show that the proposed face representation approach is more effective and robust to the in-depth rotation of pose when there are only frontal images in the training set.

  • Hilbert Scan based Tree Representation for Image Search

    Pengyi Hao, Sei-ichiro Kamata

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     499 - 504  2010年  [査読有り]

     概要を見る

    In this paper, Hilbert scan based tree representation (HSBT) is presented for image search. Unlike common ways decreasing the number of interest points or reducing the dimensions of features or using searching methods to match interest points, the proposed method builds a tree for each image and gives a new distance measure to calculate the similarity between the query and images in database. In the proposed approach, Hilbert scan for arbitrarily-sized arrays is used to map the interest points from two-dimensional space to one-dimensional space at first. Then, interest points set is divided into several parts by a separation way, and a grouping strategy is given to build a tree for each image. Experimental results show that the proposed approach is space saving. That is because it only stores clustering center and relevant information of each node in the tree. It is also time saving since the similarity calculation is up to the nodes of tree rather than all the descriptors of image. At the same time, the retrieval precision is good, because Hilbert scanning preserves the correlation in two-dimensional image, so nodes of tree are shaped according to the compactness of interest points which can employ the local information as much as possible.

  • Fast Polar Harmonic Transforms

    Zhuo Yang, Sei-ichiro Kamata

    11TH INTERNATIONAL CONFERENCE ON CONTROL, AUTOMATION, ROBOTICS AND VISION (ICARCV 2010)     673 - 677  2010年  [査読有り]

     概要を見る

    Polar Harmonic Transform (PHT) is termed to represent a set of transforms those kernels are basic waves and harmonic in nature. PHTs consist of Polar Complex Exponential Transform (PCET), Polar Cosine Transform (PCT) and Polar Sine Transform (PST). They are proposed to represent invariant image patterns for two dimensional image retrieval and pattern recognition tasks. They are demonstrated to show superiorities comparing with other methods on describing rotation invariant patterns for images. Kernel computation of PHTs is also simple and has no numerical stability issue. However in order to increase the computation speed, fast computation method is needed especially for real world applications like limited computing environments, large image databases and realtime systems. This paper presents Fast Polar Harmonic Transforms (FPHTs) including Fast Polar Complex Exponential Transform (FPCET), Fast Polar Cosine Transform (FPCT) and Fast Polar Sine Transform (FPST) that are deduced based on mathematical properties of trigonometric functions. The proposed FPHTs are averagely over 6 similar to 8 times faster than PHTs that significantly boost computation process. The experimental results on both synthetic and real data are given to illustrate the effectiveness of the proposed fast transforms.

    DOI

    Scopus

    11
    被引用数
    (Scopus)
  • A Novel Face Representation Toward Pose Invariant Face Recognition

    Liang Yu, Sei-ichiro Kamata, Yong Fang

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     179 - 183  2010年  [査読有り]

     概要を見る

    Though elastic bunch graph matching (EBGM) has a good performance on face recognition in the distortion of facial expression, it is still not robust enough to in-depth rotation. To solve this problem, a novel face representation approach based on the space-filling tree is proposed in this paper. This kind of representation shows a better performance than Elastic bunch graph matching (EBGM) in in-depth rotation of pose especially when there are only frontal images in the training set. With the proposed face representation approach, the face recognition system is built. Experimental results on the FERET standard database show that the proposed face representation approach is more effective and robust to the in-depth rotation of pose when there are only frontal images in the training set.

    DOI

    Scopus

  • Face Recognition with Local Gradient Derivative Patterns

    Xianchun Zheng, Sei-ichiro Kamata, Liang Yu

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     667 - 670  2010年  [査読有り]

     概要を見る

    In this work, we present a novel local pattern descriptor, Local Gradient Derivative Pattern (LGDP) to face recognition which considers more detailed information than the Local Binary Pattern (LBP). The face image is first divided into several small regions from which Local Gradient Derivative Pattern (LGDP) histograms are extracted and concatenated into a single, spatially enhanced feature vector to be used as a face descriptor. Three well-known and challenge-ORL, Yale and FERET face databases are used in the performances to evaluate the method. The experiments result clearly show that the proposed method give us a better performance than some other methods.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • 3D Reconstruction from a Single Image for a Chinese Talking Face

    Ning Liu, Ning Fang, Seiichiro Kamata

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     1613 - 1616  2010年  [査読有り]

     概要を見る

    This paper proposes an automatic 3D reconstruction approach for a Chinese talking face by a generic model and a single image. Firstly, an improved color-based ASM method is used to detect the face area and get the 2D face feature points automatically from the given image, which is not restricted to full frontal one. Then, color information is used to correct the location of face feature points. Finally, after text mapping, a particular and realistic 3D face model is deformed from a generic model. Using ASM face feature points extraction and correction based on skin color model, the problem of side face information missing is successfully resolved. Depending on only one image and one generic model, the computing cost of memory and time is largely reduced. The 3D face reconstructed can be easily deformed to form different expressions and mouth shapes. Experiments show that this approach is fast and efficient and has an output of a lifelike Chinese talking face.

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • Document Layout Analysis and Reading Order Determination for a Reading Robot

    Yucun Pan, Qunfei Zhao, Seiichiro Kamata

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     1607 - 1612  2010年  [査読有り]

     概要を見る

    In this paper an efficient approach of document layout analysis and reading order determination is proposed for a reading robot. Firstly the input document images are preprocessed to remove noises, connect lines and domains, and to reduce the computation time. Secondly a bottom-up, parameter-independent, two-step layout analysis algorithm based on morphology is used, which outlines the geometry of the maximum homogeneous regions and classifies them into texts, tables, and pictures. Finally the reading order is determined, by a top-down recursive hierarchy algorithm derived from XY-cut, using a set of rules depending on layout information. Important parameters are acquired using statistic information of the given images to adapt to different types of documents. The proposed algorithm is applied to a large number of document images and the experimental results show that it makes the reading robot be able to read paper documents of different languages, even with complex layout structure.

    DOI

    Scopus

    11
    被引用数
    (Scopus)
  • Hilbert Scan based Tree Representation for Image Search

    Pengyi Hao, Sei-ichiro Kamata

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     499 - 504  2010年  [査読有り]

     概要を見る

    In this paper, Hilbert scan based tree representation (HSBT) is presented for image search. Unlike common ways decreasing the number of interest points or reducing the dimensions of features or using searching methods to match interest points, the proposed method builds a tree for each image and gives a new distance measure to calculate the similarity between the query and images in database. In the proposed approach, Hilbert scan for arbitrarily-sized arrays is used to map the interest points from two-dimensional space to one-dimensional space at first. Then, interest points set is divided into several parts by a separation way, and a grouping strategy is given to build a tree for each image. Experimental results show that the proposed approach is space saving. That is because it only stores clustering center and relevant information of each node in the tree. It is also time saving since the similarity calculation is up to the nodes of tree rather than all the descriptors of image. At the same time, the retrieval precision is good, because Hilbert scanning preserves the correlation in two-dimensional image, so nodes of tree are shaped according to the compactness of interest points which can employ the local information as much as possible.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Face Detection in Color Images Based on Skin Color Models

    Li Zou, Sei-ichiro Kamata

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE     681 - 686  2010年  [査読有り]

     概要を見る

    Face finding is a very important initial step towards building up a fully automated face recognition system. Face detection by detecting skin like colors can achieve a high detection rate. In this paper, we presented a novel algorithm for face detection in color images with complex backgrounds. First a parallel structure for skin color detection is proposed to improve the accuracy of detections. The concept of the probability image has been introduced to utilize the color information in the traditional face detection methods specific for gray-scale images. After that, a classifier obtained from Adaboost training is applied to the result of skin detection to reduce the false positives. An experiment has been implemented to verify the improvement of this proposed research. And the proposed approach achieved a better result in this experiment.

    DOI

    Scopus

    12
    被引用数
    (Scopus)
  • Adaptive histogram analysis for image enhancement

    Qieshi Zhang, Hiroshi Inaba, Sei-Ichiro Kamata

    Proceedings - 4th Pacific-Rim Symposium on Image and Video Technology, PSIVT 2010     408 - 413  2010年

     概要を見る

    One image processing application is to reconstruct the original scene from the low quality images. Considering the idea histogram distribution can reflect good vision effect. So many histogram analyzing based methods have been studied recently. However, some methods require users to set some parameters or condition, and cannot get the optimal results automatically. To overcome those short come, this paper presents an Adaptive Histogram Separation and Mapping (AHSM) method for Backlight image enhancement. First, we separate the histogram by binary tree structure with the proposed Adaptive Histogram Separation Unit (AHSU). And then mapping the Low Dynamic Range (LDR) histogram partition into High Dynamic Range (HDR). By doing this, the excessive or scarcity enhancement can be avoid. The experimental results show that the proposed method can gives better enhancement results, also compared with some histogram analyzing based methods and get better results. © 2010 IEEE.

    DOI

    Scopus

    7
    被引用数
    (Scopus)
  • 3D OBJECT MATCHING BASED ON SPHERICAL HILBERT SCANNING

    Can Tong, Sei-ichiro Kamata

    2010 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING     2941 - 2944  2010年  [査読有り]

     概要を見る

    This paper describes a novel method to match objects in cluttered scenes. This method makes use of Hilbert scanning of feature points in Hough space. We use a 3D Hough transform to obtain a spectrum on which 3D features are concentrated on the sphere. Then, based on the obtained Hough Spectrum, we apply Hilbert scanning on the sphere to match the objects. Using this approach, we can match the object correctly and robustly in both overlapping and noise situation. The characteristic of this method is that it is a global matching method without an estimate of the rotation first and suffering from computational complexity brought by voting/correlation procedure. The experiment results show that the method is more effective compared to existing methods in both matching rate and robustness.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • IMAGE DESCRIPTION WITH 1D LOCAL PATTERNS BY MULTI-SCANS: AN APPLICATION TO FACE RECOGNITION

    Wei Zhou, Alireza Ahrary, Sei-ichiro Kamata

    2010 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING     4553 - 4556  2010年  [査読有り]

     概要を見る

    In this work, a novel approach which called 1D local patterns by multi-scans(1DLPMS) for presenting the local features is proposed and its simplifications and extensions to facial image analysis are also considered. First, multi-scans are applied to capture different spatial information on the image with less computation than some traditional ways, such as Local Binary Patterns(LBP). Then, some 1D local patterns are given to encode the local features based on different coding rules. To make the proposed approach computationally simpler and easy to extend, grouped 1D local patterns by multiscans(G1DLPMS) is studied, which divides 1DLPMS into several groups and uses the co-occurrences of these groups. Performance assessment in face recognition under different challenges shows that the proposed approach is superior than traditional ones.

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • Fast Polar Harmonic Transforms

    Zhuo Yang, Sei-ichiro Kamata

    11TH INTERNATIONAL CONFERENCE ON CONTROL, AUTOMATION, ROBOTICS AND VISION (ICARCV 2010)     673 - 677  2010年  [査読有り]

     概要を見る

    Polar Harmonic Transform (PHT) is termed to represent a set of transforms those kernels are basic waves and harmonic in nature. PHTs consist of Polar Complex Exponential Transform (PCET), Polar Cosine Transform (PCT) and Polar Sine Transform (PST). They are proposed to represent invariant image patterns for two dimensional image retrieval and pattern recognition tasks. They are demonstrated to show superiorities comparing with other methods on describing rotation invariant patterns for images. Kernel computation of PHTs is also simple and has no numerical stability issue. However in order to increase the computation speed, fast computation method is needed especially for real world applications like limited computing environments, large image databases and realtime systems. This paper presents Fast Polar Harmonic Transforms (FPHTs) including Fast Polar Complex Exponential Transform (FPCET), Fast Polar Cosine Transform (FPCT) and Fast Polar Sine Transform (FPST) that are deduced based on mathematical properties of trigonometric functions. The proposed FPHTs are averagely over 6 similar to 8 times faster than PHTs that significantly boost computation process. The experimental results on both synthetic and real data are given to illustrate the effectiveness of the proposed fast transforms.

  • 荒天時の車載カメラ映像におけるフロントガラスへの付着物による前方遮蔽部の修繕に関する一考察(高精細画像の処理・表示,及び一般)

    稲葉 洋, 鎌田 清一郎

    電子情報通信学会技術研究報告. IE, 画像工学   109 ( 292 ) 33 - 38  2009年11月

     概要を見る

    車載カメラは車両周辺の環境を入力するためのセンサとして広く用いられる.しかしながら,センサとしてのカメラは晴天時では有効に働くものの,荒天時ではレンズへの付着物により画像の有効部分が減り性能低下が懸念される.同様の状況は,ドライブレコーダのようなカメラを有する車内設置型の装置において,車両のガラスを通して車外を撮影する場合にも起こりうる.本研究では,荒天時の車載カメラ映像に対する視認性改善に向けた基礎的検討として後者の状況に着目し,フロントガラスへの付着物により前方が遮蔽された領域の修繕を試みる.手法は,付着物として雨滴を想定し,一台の一般的なカメラを用いて雨滴の検出,および,その部分の修繕を行い,雨滴が除去された映像を生成する.各時刻における画像の修繕は,直前の数フレーム分の画像を用い,それらから雨滴の存在しない部分を統合して行う.本文では一台のカメラの映像を用いて修繕を行う際の問題点について考察し,実験においていくつかの修繕例を示す.

    DOI CiNii

  • 局所的指標による予測器選択を用いた可逆画像圧縮(研究速報,画像符号化,<特集>画像符号化・映像メディア処理レター)

    杉本 憲治郎, 黒木 祥光, 鎌田 清一郎

    電子情報通信学会論文誌. D, 情報・システム   92 ( 10 ) 1698 - 1701  2009年10月

     概要を見る

    予測符号化において,隣接画素間の差分(こう配)に基づき予測器を選択する手法が提案されている.しかしながらこう配と予測器間の関係についての議論は定性的な場合が多い.本論文では,こう配の代わりに平均予測値との差分を局所的指標として用い,従来手法における予測器との関係を定量的に議論する.また局所的指標を用いた予測手法及びスキームを提案し,その予測効率と符号量を測定する.

    CiNii

  • 局所的指標による予測器選択を用いた可逆画像圧縮

    杉本憲治郎, 黒木祥光, 鎌田清一郎

    電子情報通信学会論文誌(D)   J92-D ( 10 ) 1698 - 1701  2009年10月

     概要を見る

    予測符号化において,隣接画素間の差分(こう配)に基づき予測器を選択する手法が提案されている.しかしながらこう配と予測器間の関係についての議論は定性的な場合が多い.本論文では,こう配の代わりに平均予測値との差分を局所的指標として用い,従来手法における予測器との関係を定量的に議論する.また局所的指標を用いた予測手法及びスキームを提案し,その予測効率と符号量を測定する.

    CiNii

  • Face Detection and Tracking in Color images Using Color Centroids Segmentation

    Qieshi Zhang, Sei-ichiro Kamata, Jun Zhang

    2008 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND BIOMIMETICS, VOLS 1-4     1008 - 1013  2009年  [査読有り]

     概要を見る

    Human face detection plays an important role in many application areas such as video surveillance, human computer interface, face recognition, face search and face image database management etc. In human face detection applications, face region usually form an inconsequential part of images. Consequently, preliminary segmentation of images into regions that contain "non-face" objects and regions that may contain "face" candidates can greatly accelerate the process of human face detection. Color information based methods take a great attention, because colors have obviously character and robust visual cue for detection. This paper proposed a new method based on RGB color centroids segmentation (CCS) for face detection. This paper include two parts, first part is color image thresholding based on CCS and the second part is face detection based on region growing and facial features structure character combined method. The experimental results show the ideal thresholding result and better than the result of other color space analysis based thresholding methods. Proposed method can conquer the influence of different background conditions, position, scale instance and orientation in images from several photo collections and database; the effect is also better than existing skin color segmentation based methods.

    DOI

    Scopus

    18
    被引用数
    (Scopus)
  • パラレル順次走査に基づく領域抽出のための新色記述

    Yang Zhuo, アハラリ アリレザ, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2009 ( 0 ) 530 - 530  2009年

     概要を見る

    A color descriptor is a numeric quantity that describes a color feature of an image. Color can be extracted from the image as a whole, a global characterization; or separately from different regions, producing a local characterization. One of the drawbacks of extracting color histograms globally is that it does not take into account the spatial distribution of the color across different areas of the image. In this paper a new color descriptor based on parallel progressive scan is proposed that solve real world problems in region detection for color images.

    CiNii

  • Single Medicine Recognition using Color Histogram

    Cai Qi, アハラリ アリレザ, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2009 ( 0 ) 529 - 529  2009年

     概要を見る

    In this paper, a new method using color histogram is proposed for recognizing printed character of single medicine. In different brightness condition, the three-dimensional color histogram also changes and a naive three-dimensional partition of color space often supports poor indexing. To circumvent this problem, image RGBs are mapped to brightness-independent chromaticity prior to indexing. Then, chi-square is used for measuring the similarity between the sample medicines and testing medicine. The experimental results show that the proposed method gives an acceptable result.

    CiNii

  • Fast Facial Feature Point Detection for 3D Face Recognition

    童 燦, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2009 ( 0 ) 528 - 528  2009年

     概要を見る

    A face recognition system that utilizes three-dimensional shape information is more robust to arbitrary view, lighting, and facial appearance. The main problem in 3D face recognition is how to detect the feature points correctly and efficiently. In this paper, we present a novel method to detect feature points for 3D face recognition. The experiment result shows that the proposed method performs better in both accuracy and efficiency than other methods.

    CiNii

  • Face Recognition with Multi-Scan and Histogram Spatially constrained Earth Mover's Distance

    周 蔚, アハラリ アリレザ, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2009 ( 0 ) 527 - 527  2009年

     概要を見る

    In this work, a multi-scan is applied for each block in the facial image while entropy is used for selecting the best scan order in that block. Based on that scan order, local feature pattern is proposed to obtain feature histograms in the face. Then, a novel matching method called Histogram Spatially constrained Earth Mover's Distance is proposed to take alignment of face into account. The experimental results show that the proposed method has higher accuracy than some other classic methods.

    CiNii

  • 接近する人物に対しての歩容認証の一検討

    萩尾 和也, アハラリ アリレザ, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2009 ( 0 ) 408 - 408  2009年

     概要を見る

    人物の歩行特徴から個人を特定する歩容認証は,被験者にかかる負担が小さな生体認証の一つとして近年注目を集めている.しかし,多くの研究では,カメラから十分に離れた人物が側面から撮影されていることを想定しており,歩行方向やカメラの視線方向,被験者とカメラ間の距離の変化の及ぼす影響については十分に議論されていない.そこで本研究では,独自の撮影環境を設営し,カメラに対して被験者が接近してくる場合の歩容認証について検討する.また,撮影にはステレオカメラを用い,歩容認証における立体情報の有用性ついても検討する.この研究の成果は,建物での入退室管理への応用などに活かすことができると考えられる.

    CiNii

  • Image Registration Based on Genetic Algorithm and Weighted Feature Correspondences

    Zhi Yuan, Alireza Ahrary, Peimin Yan, Sei-ichiro Kamata

    ISCE: 2009 IEEE 13TH INTERNATIONAL SYMPOSIUM ON CONSUMER ELECTRONICS, VOLS 1 AND 2     770 - +  2009年  [査読有り]

     概要を見る

    Super resolution is a technique of enhancing image resolution by combining information from multiple images. It is widely applied in fields like camera surveillance, satellite imaging, pattern recognition, etc. One challenging problem of super resolution is its high demand on image registration accuracy. This paper introduces a high accuracy registration approach for the purpose of super resolution. It is invariant to translation, scaling, rotation, and noise, and can be used to automatically obtain the Maximize a Likelihood Estimation (MLE) of image homography (registration result) using information only contained within the images themselves. An effective Genetic Algorithm based approach is used to filter out all the mismatches. Comparison with RANSAC and Keren's method will be given to prove the effectiveness of the proposed method.

  • Face Recognition using Local Quaternion Patters and Weighted Spatially constrained Earth Mover's Distance

    Wei Zhou, Alireza Ahrary, Sei-ichiro Kamata

    ISCE: 2009 IEEE 13TH INTERNATIONAL SYMPOSIUM ON CONSUMER ELECTRONICS, VOLS 1 AND 2     648 - +  2009年  [査読有り]

     概要を見る

    This paper presents a novel algorithm for face recognition. Local Quaternion Patters (LQP) is proposed for presenting the feature parts in the face. To keep the spatial feature of the face, an asymmetric similarity measure Weighted Spatially constrained Earth Mover's Distance (WSEMD) is studied for classification. In this step, the source image is partitioned into non overlapping local patches while the destination image is represented as a set of overlapping local patches at different positions and Gaussian Kernel is used. Finally, local and global weighting is applied to get a more accurate classifier. To evaluate the proposed method and its performance, three well-known and challenge face databases - ORL, Yale and FERET are used in our study. The experimental results show that the proposed method has higher accuracy than some other classic methods.

  • On-Line Signature Matching Based on Hilbert Scanning Patterns

    Alireza Ahrary, Hui-ju Chiang, Sei-ichiro Kamata

    ADVANCES IN BIOMETRICS   5558   1190 - +  2009年  [査読有り]

     概要を見る

    Signature verification is a challenging task, because only a small set of genuine samples call be acquired and usually no forgeries are available in real application. In this paper, we propose a novel approach based oil Hilbert scanning patterns and Gaussian mixture models for automatic on-line signature verification. Our system is composed of a similarity measure based oil Hilbert scanning patterns and a simplified Gaussian mixture model for decision-level evaluation. To be practical, we introduce specific simplification strategies for model building and training. The system is compared to other state-of-the-art systems based oil the results of the First International Signature Verification Competition (SVC 2004). Experiments are conducted to verify, the effectiveness of our system.

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • An Adaptive Tone Mapping Algorithm for High Dynamic Range Images

    Jian Zhang, Sei-ichro Kamata

    COMPUTATIONAL COLOR IMAGING   5646   207 - 215  2009年  [査読有り]

     概要を見る

    Real world scenes contain a large range of light intensities which range from dim starlight to bright; sunlight. A common task of tone mapping algorithms is to reproduce high dynamic range(HDR) images on low dynamic range(LDR) display devices such as printers and monitors. In this paper, a new tone mapping algorithm is proposed for the display of HDR images. Inspired by the adaptive process of the human visual system, the proposed algorithm utilized the center-surround Retinex processing. The novelty of our method is that the local details are enhanced according to a non-linear adaptive spatial filter (Gaussian filter), whose shape is adapted to high-contrast edges of the image. The proposed method uses an adaptive surround instead of the traditional pre-defined circular. Therefore, the algorithm can preserve visibility and contrast impression of high dynamic range scenes in the common display devices. The proposed method is tested on a variety of HDR, images, and we also compare it to previous work. The results show good performance of our method in terms of visual quality.

    DOI

    Scopus

    4
    被引用数
    (Scopus)
  • Image registration based on genetic algorithm and weighted feature correspondences

    Zhi Yuan, Alireza Ahrary, Peimin Yan, Sei-Ichiro Kamata

    Digest of Technical Papers - IEEE International Conference on Consumer Electronics     42 - 46  2009年

     概要を見る

    Super resolution is a technique of enhancing image resolution by combining information from multiple images. It is widely applied in fields like camera surveillance, satellite imaging, pattern recognition, etc. One challenging problem of super resolution is its high demand on image registration accuracy. This paper introduces a high accuracy registration approach for the purpose of super resolution. It is invariant to translation, scaling, rotation, and noise, and can be used to automatically obtain the Maximize a Likelihood Estimation (MLE) of image homography (registration result) using information only contained within the images themselves. An effective Genetic Algorithm based approach is used to filter out all the mismatches. Comparison with RANSAC and Keren's method will be given to prove the effectiveness of the proposed method. ©2009 IEEE.

    DOI

    Scopus

    4
    被引用数
    (Scopus)
  • A new on-line signature verification algorithm using hilbert scanning patterns

    Alireza Ahrary, Sei-Ichro Kamata

    Digest of Technical Papers - IEEE International Conference on Consumer Electronics     276 - 279  2009年

     概要を見る

    Signature verification is a challenging task, because only a small set of genuine samples can be acquired and usually no forgeries are available in real application. In this paper, we propose a new function-based method with Hilbert Scanning patterns for automatic on-line signature verification. The proposed method is compared to other state-of-the-art methods based on the results of the First International Signature Verification Competition (SVC 2004). Experiments are conducted to verify the effectiveness of our method. ©2009 IEEE.

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • 5-3 図書に記された文字の認識と読み上げ(第5部門 ヒューマンインフォメーション2)

    小松原 幸弘, 山内 幸治, 鎌田 清一郎

    映像情報メディア学会冬季大会講演予稿集   2009 ( 0 ) _5 - 3-1_  2009年

     概要を見る

    In this study, we try to read printed characters on the book. This paper describes the method to segment a sentence to characters and search a pronunciation of the character by the dictionary.

    DOI CiNii

  • Face recognition with local feature patterns and histogram spatially bonstrained earth mover's distance

    Wei Zhou, Alireza Ahrary, Sei-Ichiro Kamata

    ICSIPA09 - 2009 IEEE International Conference on Signal and Image Processing Applications, Conference Proceedings     374 - 379  2009年

     概要を見る

    In this work, two novel local feature patterns-Modified Local Binary patterns (MLBP) and local Ternary patterns (LIP), are proposed for extract features in the facial image, which use some distinct rule to code the values in a label, respectively. These patterns are more invariant to illuminance and face expression compared to traditional one. After getting the local feature patterns, in order to take alignment of face into account, a novel matching method called Histogram Spatially constrained Earth Mover's Distance(HSEMD) is proposed. In this step, the source image is partitioned into non-overlapping local regions while the destination image is represented as a set of overlapping local regions at different positions. Meanwhile, multi-scale cascade mechanism is studied for extracting more feature patterns and obtaining global information of the face.The performance of the proposed method is assessed in the face recognition problem under different challenges. The experimental results show that the proposed method has higher accuracy than some other classic methods.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • Linear Predictor Using 3-D Projection for Video Loss less Compression

    Daejung Bang, Haijiang Tang, Sei-ichiro Kamata

    ISIE: 2009 IEEE INTERNATIONAL SYMPOSIUM ON INDUSTRIAL ELECTRONICS     1897 - 1901  2009年  [査読有り]

     概要を見る

    Recently, video lossless compression has been developed for applying it to digital cinema, video archiving of contents, etc. Video lossless compression is important in image processing problem since a large image requires a large amount of storage space. The purpose of this paper is to enhance the predictor used for the lossless compression of video. In this paper, we propose the 3-dimensional predictor for the effective prediction. In addition, the three-dimensional spatio-temporal gradient is adopted to improve the conventional image compression methods such as GAP, MED which are two-dimensional predictions based on horizontal and vertical gradients. The spatio-temporal gradient is a spatial data resulted from the projection of triangular prism composed of the neighborhood pixels. From the experimental results compared with the previous prediction methods, we confirmed that the prediction using proposed method is more efficient.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • NIR: Content Based Image Retrieval on Cloud Computing

    Zhuo Yang, Sei-ichiro Kamata, Alireza Ahrary

    2009 IEEE INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTING AND INTELLIGENT SYSTEMS, PROCEEDINGS, VOL 3   3   556 - +  2009年  [査読有り]

     概要を見る

    NIR is an open source cloud computing 'enabled content based Image retrieval system With the development and popularization of cloud computing, more and more researchers from different research areas do research with the help of cloud computing Nowadays content based image retrieval as one of the challenging and emerging technologies is high computation task because of the algorithm computation complexity and big amount of data As based on cloud computing infrastructure, NIR is easy to extent and flexible for deployment As an open source project, NIR can be improved on demand and integrated to other existing systems This paper presents our ideas, findings, design and the system from our work of NIR

    DOI

    Scopus

    22
    被引用数
    (Scopus)
  • 3D Face Recognition Based on Fast Feature Detection and Non-rigid Iterative Closest Point

    Can Tong, Sei-ichiro Kamata, Alireza Ahrary

    2009 IEEE INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTING AND INTELLIGENT SYSTEMS, PROCEEDINGS, VOL 4   4   509 - +  2009年  [査読有り]

     概要を見る

    This paper presents a 3D face recognition algorithm using fast landmark detection and non-rigid iterative Closest Point (ICP) algorithm. The proposed approach can estimate the facial feature region using the anthropometric face model after pose correction, and accurately detect 9 facial landmarks (nose tip, sellion, inner and outer eye corners, nostrils and mouth center). An extension of ICP algorithm has also been proposed to matching the non-rigid 3D face shapes. Experimental results demonstrate that compared to the existing methods, the proposed approach can efficiently detect human facial landmarks and satisfactorily deal with the 3D face matching problem.

    DOI

    Scopus

  • Fingerprint Image Enhancement by Super Resolution with Early Stopping

    Zhi Yuan, Jiong Wu, Sei-ichiro Kamata, Alireza Ahrary, Peimin Yan

    2009 IEEE INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTING AND INTELLIGENT SYSTEMS, PROCEEDINGS, VOL 4   4   527 - +  2009年  [査読有り]

     概要を見る

    This paper addressed the problem of multi-frame image super resolution and its implementation to fingerprint image.,The use of computers in fingerprints recognition is highly desirable in many applications where security is an important concern. However, as the performance of fingerprint recognition algorithm relies heavily on the quality of the input fingerprint images, it is essential to incorporate a fingerprint enhancement algorithm in the minutiae extraction module. Typical approaches use Gabor filter to raise the contrast between fingerprint ridges and valleys. But few of them try to increase the very basic resolution of fingerprint image. Therefore, we proposed a computer vision solution to this problem which involves the technique of super resolution. This approach can be incorporated into any traditional fingerprint enhancement algorithm as a preprocessing step, rendering a better overall enhancement result.

    DOI

    Scopus

    8
    被引用数
    (Scopus)
  • An automatic image-map alignment algorithm based on Mutual Information and Hilbert scan

    Tian, Li, Kamata, Sei Ichiro

    European Signal Processing Conference    2008年12月

     概要を見る

    An algorithm for automatic image-map alignment problem using a new similarity measure named Edge-Based Code Mutual Information (EBCMI) and Hilbert scan is presented in this study. Because image and map are very different in their representations, the normal Mutual Information (MI) using the intensity in traditional alignment method may result in misalignment. To solve the problem, codes which are robust to the differences between the image-map pairs are constructed and Mutual Information of the codes is computed as the similarity measure for the alignment. We convert the 3-D transformation search space in alignment to a 1-D search space sequence by using 3-D Hilbert Scan. A new search strategy is also proposed on the 1-D search space sequence. The experimental results show that the proposed EBCMI outperformed the normal MI and some other similarity measures and the proposed search strategy gives flexibility between efficiency and accuracy for automatic imagemap alignment task.

  • 動きベクトルを用いた車載カメラ映像に含まれる雨滴の抽出に関する一考察(高精細度画像処理・表示及び一般)

    稲葉 洋, 鎌田 清一郎

    電子情報通信学会技術研究報告. IE, 画像工学   108 ( 324 ) 59 - 63  2008年11月

     概要を見る

    本研究では,降雨時の車載カメラ映像に含まれる雨滴により生じる視界不良部の抽出を試みる.提案手法は,映像の隣接フレームにおける画素の動きベクトルを用い,各画素における過去の動きベクトルの解析に基づき雨滴を抽出するものである.本手法を,降雨時の車載カメラ映像1例に適用した結果,4割程度の雨滴を抽出した.具体的には,短時間において,雨滴に背景全体が写り込み,かつ,雨滴の大きさが小さい場合抽出が行え,雨滴に近距離の前方が写り込む,また,雨滴がガラス上部にある場合抽出が困難であった.

    DOI CiNii

  • 平均予測値との差分による予測器選択を用いた可逆画像圧縮(高精細度画像処理・表示及び一般)

    杉本 憲治郎, 黒木 祥光, 鎌田 清一郎

    電子情報通信学会技術研究報告. IE, 画像工学   108 ( 324 ) 65 - 69  2008年11月

     概要を見る

    予測符号化において,局所的な指標に基づき予測器を切り替えることで高効率化を狙う手法が数多く提案されている.局所的な指標として隣接画素間の差分(勾配)を用いた手法がMEDやGAPをはじめ多くあるが,勾配と予測器間の関係についての定量的な議論は少ない.本稿では局所的な指標として平均予測値との差分を用い,その指標と予測器間の関係の定量的な議論を試みる.またその議論を基に設計された予測手法およびスキームを提案する.性能比較実験では,提案予測手法はGAPに比べ予測誤差エントロピを0.070[bits/pixel]減少でき,提案スキームはCALICに比べて平均符号長を0.016[bits/pixel]削減できた.

    CiNii

  • A Two-Stage Point Pattern Matching Algorithm Using Ellipse Fitting and Dual Hilbert Scans

    Li Tian, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E91D ( 10 ) 2477 - 2484  2008年10月  [査読有り]

     概要を見る

    Point Pattern Matching (PPM) is an essential problem in many image analysis and computer vision tasks. This paper presents a two-stage algorithm for PPM problem using ellipse fitting and dual Hilbert scans. In the first matching stage, transformation parameters, are coarsely estimated by using four node points of ellipses which are fitted by Weighted Least Square Fitting (WLSF). Then, Hilbert scans are used in two aspects of the second matching stage: it is applied to the similarity measure and it is also used for search space reduction. The similarity measure named Hilbert Scanning Distance (HSD) can be computed fast by converting the 2-D coordinates of 2-D points into 1-D space information using Hilbert scan. On the other hand, the N-D search space can be converted to a 1-D search space sequence by N-D Hilbert Scan and art efficient search strategy is proposed on the 1-D search space sequence. In the experiments. we use both simulated point set data and real fingerprint images, to evaluate the performance of our algorithm and our algorithm gives satisfying results both in accuracy and efficiency.

    DOI

    Scopus

    3
    被引用数
    (Scopus)
  • 18-11 Online signature matching based on Hilbert-Scanning patterns

    CHIANG Huiju, ZHANG Jian, AHRARY Alireza, KAMATA Seiichiro

    映像情報メディア学会年次大会講演予稿集   ( 2008 ) "18 - 11-1"-"18-11-2"  2008年08月

     概要を見る

    Signature-based personal identification systems are used and accepted widely due to its distinctness and stability. In this paper, we propose a new function-based method with Hilbert-Scanning patterns for signature matching. We evaluate the performance by using the online signature database Signature Verification Competition (SVC) 2004.

    CiNii

  • 18-6 動画像可逆圧縮のための時空間勾配を用いた3次元予測(第18部門 符号化・セキュリティ)

    房 大政, 唐 海江, 鎌田 清一郎

    映像情報メディア学会年次大会講演予稿集   ( 2008 ) "18 - 6-1"-"18-6-2"  2008年08月

     概要を見る

    In this paper, we propose video lossless compression based on 3-dimensional prediction using spatio-temporal gradients. The proposed method predicted the target pixel using the LGM(Local Gradient Magnitude), which is the gradient between the neighborhood pixel and the current pixel in the spatio-temporal. From the experimental results, we demonstrated that the spatio-temporal gradients prediction is more efficient.

    CiNii

  • D-12-109 点パターンマッチングのための二段階マッチングアルゴリズム(D-12. パターン認識・メディア理解,一般セッション)

    田 黎, 鎌田 清一郎

    電子情報通信学会総合大会講演論文集   2008 ( 2 ) 240 - 240  2008年03月

    CiNii

  • D-11-55 可逆画像圧縮のためのメディアン適応予測の改善(D-11. 画像工学,一般セッション)

    杉本 憲治郎, 黒木 祥光, 鎌田 清一郎

    電子情報通信学会総合大会講演論文集   2008 ( 2 ) 55 - 55  2008年03月

    CiNii

  • TK-2-4 空間充填曲線による画像圧縮検索(TK-2. 北九州での知的クラスター創成事業(第1期)の概要と成果・課題と展望,大会委員会企画)

    鎌田 清一郎

    電子情報通信学会総合大会講演論文集   2008 ( 1 ) "SSS - 6"-"SSS-7"  2008年03月

    CiNii

  • An N-dimensional Pseudo-Hilbert scan for arbitrarily-sized hypercuboids

    Jian Zhang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES   E91A ( 3 ) 846 - 858  2008年03月  [査読有り]

     概要を見る

    The N-dimensional (N-D) Hilbert curve is a one-to-one mapping between N-D space and one-dimensional (I-D) space. It is studied actively in the area of digital image processing as a scan technique (Hilbert scan) because of its property of preserving the spatial relationship of the N-D patterns. Currently there exist several Hilbert scan algorithms. However. these algorithms have two strict restrictions in implementation. First, recursive functions are used to generate a Hilbert curve, which makes the algorithms complex and computationally expensive. Second, all the sides of the scanned region must have the same size and the length must be a power of two, which limits the application of the Hilbert scan greatly. Thus in order to remove these constraints and improve the Hilbert scan for general application, a nonrecursive N-D Pseudo-Hilbert scan algorithm based on two look-up tables is proposed in this paper. The merit of the proposed algorithm is that implementation is much easier than the original one while preserving the original characteristics. The experimental results indicate that the Pseudo-Hilbert scan can preserve point neighborhoods as much as possible and take advantage of the high correlation between neighboring lattice points, and it also shows the competitive performance of the Pseudo-Hilbert scan in comparison with other common scan techniques. We believe that this novel scan technique undoubtedly leads to many new applications in those areas can benefit from reducing the dimensionality of the problem.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • Fingerprint Matching Using Dual Hilbert Scans

    Li Tian, Liang Chen, Sei-ichiro Kamata

    SITIS 2007: PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON SIGNAL IMAGE TECHNOLOGIES & INTERNET BASED SYSTEMS   3 ( 1 ) 593 - 600  2008年

     概要を見る

    A new fingerprint matching algorithm using dual Hilbert scans is presented in this study. We treat the fingerprint matching as point pattern matching problem and Hilbert scans are used in two aspects of the matching problem: one is applied to the similarity measure and the other is used in search space reduction. The similarity measure named Hilbert Scanning Distance (HSD) can be computed fast by converting the 2-D coordinates of 2-D images into 1-D space information using Hilbert scan. On the other hand, the 3-D search space can be converted to a 1-D search space sequence. The proposed method has been tested on FVC2002 database. The experimental results show that our method can implement fingerprint matching robustly and efficiently. The performance evaluation EER (Equal-Error Rate) generally used is very low by our algorithm.

    DOI

    Scopus

    4
    被引用数
    (Scopus)
  • POLSAR画像を用いた橋高度の測定

    李 虎栄, 王 海鵬, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2008 ( 0 ) 63 - 63  2008年

     概要を見る

    POLSARデータを用いた橋高度の測定方法を提案する。de-orientation理論と分類パラメータを用い,Pi-SARのポライメトリックデータでの橋目標の1次、2次と3次散乱の画像生成原理を分析し、画像での位置を判断する。またフィルタリングとクラスタリング処理を行い,SAR画像からそれぞれの散乱画像を抽出し、2次散乱と3次散乱の画像距離によって橋の高度を測定する。最後に日本のナルト大橋のPi-SAR画像と中国の東海大橋のALOS-PALSAR画像を利用してその高度を測定し、実際高度と比較して本手法の有効性を示す。

    CiNii

  • 色分析に基づく道路標識の自動検出方法

    Zhang Qieshi, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2008 ( 0 ) 617 - 617  2008年

     概要を見る

    Road sign detection is one of the major concerned topics in the field of driving safety and intelligent vehicle. In this paper, a novel model based on Color Barycenters Hexagon (CBH) is proposed and used to detect road sign usefully. In CBH model, full color images are calculated the color barycenters and get the barycenters region, then automatic select the idea threshold curves to separate the barycenters region of interest (ROI) aiming to detect the road sign. Because of the practically images have many noise, and the existing color space cannot separate the ROI ideally, the proposed CBH model can thresholding the principal color of ROI and have high robust. With suitably thresholding and operations, road sign on various scene images can be detected.

    CiNii

  • 分散と重み付きDPマッチングを用いた顔認識

    周 蔚, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2008 ( 0 ) 616 - 616  2008年

     概要を見る

    This paper presents a novel algorithm for face recognition. Variance is used for extracting the feature vector and then Dynamic Programming (DP) is applied for matching, since the length of each feature vector is different. At last, some weighted values are added for final recognition. These weighted values can improve the recognition rate greatly. To evaluate the proposed method and its performance, a well-known face database ORL is used in our study. The experimental results show that the proposed method is much better than other existing method, such as PCA, 2DPCA, LDA, LBP and so on.

    CiNii

  • Automatic Image-Map Alignment Using Edge-Based Code Mutual Information and 3-D Hilbert Scan

    Sei-Ichiro Kamata, Li Tian

    Journal of the Institute of Image Electronics Engineers of Japan   37 ( 3 ) 223 - 230  2008年

     概要を見る

    This study presents a new algorithm for automatic image-map alignment problem using a new similarity measure named Edge-Based Code Mutual Information (EBCMI) and 3-D Hilbert scan. In general, each image-map pair can be viewed as two special multimodal images, however, are very different in their representations such as the intensity. Therefore, the normal Mutual Information (MI) using the intensity in traditional alignment method may result in misalignment. To solve the problem, codes based on the edges of the image-map pairs are constructed and Mutual Information of the codes is computed as the similarity measure for the alignment in our method. Since Edge-Based Code (EBC) is robust to the differences between the image-map pairs in their representations, EBCMI also can overcome the differences. On the other hand, the 3-D search space in alignment can be converted to a 1-D search space sequence by 3-D Hilbert Scan and a new search strategy is proposed on the 1-D search space sequence. The experimental results show that the proposed EBCMI performed better than the normal MI and some other similarity measures and the proposed search strategy gives flexibility between efficiency and accuracy for automatic image-map alignment task. © 2008, The Institute of Image Electronics Engineers of Japan. All rights reserved.

    DOI

    Scopus

  • Image enhancement by analysis on embedded surfaces of images and a new framework for enhancement evaluation

    Li Tian, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E91-D ( 7 ) 1946 - 1954  2008年

     概要を見る

    Image enhancement plays an important role in many machine vision applications on images captured in low contrast and low illumination conditions. In this study, we propose a new method for image enhancement based on analysis on embedded surfaces of images. The proposed method gives an insight into the relationship between the image intensity and image enhancement. In our method, scaled surface area and the surface volume are proposed and used to reconstruct the image iteratively for contrast enhancement, and the illumination of the reconstructed image can also be adjusted simultaneously. On the other hand, the most common methods for measuring the quality of enhanced images areMean Square Error (MSE) or Peak Signal-to-Noise-Ratio (PSNR) in conventional works. The two measures have been recognized as inadequate ones because they do not evaluate the result in the way that the human vision system does. This paper also presents a new framework for evaluating image enhancement using both objective and subjective measures. This framework can also be used for other image quality evaluations such as denoising evaluation. We compare our enhancement method with some well-known enhancement algorithms, including wavelet and curvelet methods, using the new evaluation framework. The results show that our method can give better performance in most objective and subjective criteria than the conventional methods. Copyright © 2008 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • Automatic Road Sign Detection Method Based on Color Barycenters Hexagon Model

    Qieshi Zhang, Sei-ichiro Kamata

    19TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOLS 1-6     1114 - 1117  2008年  [査読有り]

     概要を見る

    Road sign detection is one of the major concerned topics in the field of driving safety and intelligent vehicle. In this paper, a novel model based on Color Barycenters Hexagon (CBH) is proposed and used to defect road sign usefully. In CBH model, full color images are calculated the color barycenters and get the barycenters region, then automatic select the idea threshold curves to separate the Region of Interest (ROI) of barycenters aiming to detect the road sign. Because of the practically images have many noise, and the existing color space cannot separate the ROI ideally. The proposed CBH model can thresholding the principal color of ROI and have high robust. With suitably thresholding and operations, road sign on various scene images can be detected.

  • Adaptive Local Contrast Enhancement for the Visualization of High Dynamic Range Images

    Jian Zhang, Sei-ichiro Kamata

    19TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOLS 1-6     2457 - 2460  2008年  [査読有り]

     概要を見る

    In this paper we present a new tone mapping algorithm for the display of high dynamic range images, inspired by adaptive process of the human visual system. The proposed algorithm is based on the center-surround Retinex processing. In our method, the local details are enhanced according to a non-linear adaptive spatial filter (Gaussian filter), whose shape (filter variance) is adapted to high-contrast edges of the image. Thus our method does not generate halo artifacts meanwhile preserves visibility and contrast impression of high dynamic range scenes in the common display devices. The proposed method is tested on a variety of HDR images and the results show the good performance of our method in terms of visual quality.

  • An Iterative Image Enhancement Algorithm and A New Evaluation Framework

    Li Tian, Sei-ichiro Kamata

    2008 IEEE INTERNATIONAL SYMPOSIUM ON INDUSTRIAL ELECTRONICS, VOLS 1-5     429 - 434  2008年  [査読有り]

     概要を見る

    Image enhancement is important for images captured in low contrast and low illumination conditions. In this study, we propose a new iterative algorithm for image enhancement based on analysis on embedded surfaces of images. In our method, scaled surface area and the surface volume are proposed and used to reconstruct the image iteratively for contrast enhancement, and the illumination of the reconstructed image can also be adjusted simultaneously. On the other hand, the most common methods for measuring the quality of enhanced images Mean Square Error (MSE) or Peak Signal-to-Noise-Ratio (PSNR) have been recognized as inadequate measures because they do not evaluate the result in the way that the human vision system does. This paper also presents a new framework for evaluating image enhancement using both objective and subjective measures. This framework can also be used for other image quality evaluations such as denoising evaluation. We compare our enhancement method with some well-known enhancement algorithms, including wavelet and curvelet methods, using the new evaluation framework. The results show that our method gives better performance in most objective and subjective criteria than the conventional methods.

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • Adaptive Local Contrast Enhancement for the Visualization of High Dynamic Range Images

    Jian Zhang, Sei-ichiro Kamata

    19TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOLS 1-6     2457 - 2460  2008年  [査読有り]

     概要を見る

    In this paper we present a new tone mapping algorithm for the display of high dynamic range images, inspired by adaptive process of the human visual system. The proposed algorithm is based on the center-surround Retinex processing. In our method, the local details are enhanced according to a non-linear adaptive spatial filter (Gaussian filter), whose shape (filter variance) is adapted to high-contrast edges of the image. Thus our method does not generate halo artifacts meanwhile preserves visibility and contrast impression of high dynamic range scenes in the common display devices. The proposed method is tested on a variety of HDR images and the results show the good performance of our method in terms of visual quality.

  • A simple tone mapping for high dynamic range image visualization using a Pseudo-Hilbert scan

    Zhang, Jian, Kamata, Sei Ichiro, Tian, Li

    Proceedings of IAPR Conference on Machine Vision Applications, MVA 2007     363 - 366  2007年12月

     概要を見る

    The Hilbert curve is one of space-filling curves published by G. Peano. There are several applications using this curve, such as image processing, computer graphics, etc. In this paper, we concentrate on a tone mapping technique for high dynamic range images using the Pseudo-Hilbert curve. Based on the neighbourhood property of the Pseudo-Hilbert scan, a fast and flexible tone reproduction method is proposed. The proposed new technique preserves visibility and contrast impression of high dynamic range scenes in low dynamic range reproduction devices. From the experimental results, we have confirmed that the proposed method produces good results on a variety of high dynamic range images.

  • Image contrast enhancement by analysis on embedded surfaces of images

    Tian, Li, Kamata, Sei Ichiro

    Proceedings of IAPR Conference on Machine Vision Applications, MVA 2007   10 ( 1 ) 90 - 93  2007年12月

     概要を見る

    Image contrast enhancement plays an important role in many machine vision applications. In this study, we propose a new method for edge enhancement based on analysis on embedded surfaces of images. The proposed method gives an insight into the relationship between the image intensities (also the gradients) and image contrast. In our method, scaled surface area and the surface volume are used to reconstruct the image for edge enhancement, and then the contrast of the reconstructed image is adjusted by a 'strengthen-weaken' process. Although, current method for edge enhancement such as curvelet transform can enhance the edge part, it does not provide good tonal rendition or color constancy sometimes. The experimental results show that our method can give good performance not only in edge enhancement, but also in tonal rendition and color constancy.

    CiNii

  • 時空間勾配を用いた3次元予測による動画像の可逆圧縮(高精細画像の処理・表示及び一般)

    安 昭映, 唐 海江, 鎌田 清一郎

    電子情報通信学会技術研究報告. IE, 画像工学   107 ( 358 ) 109 - 113  2007年11月

     概要を見る

    近年,放送映像などの素材蓄積,ディジタツシネマなどへの応用を目的として,動画像の可逆符号化が検討されている.本論文では,動画像を対象とし,時空間勾配を利用した3次元予測による可逆圧縮について述べる.これは,従来GAR,MEDなどの水平および垂直方向のエッジに着目した予測方式に対して,エッジの方向をより細かく捉えた,2次元空間勾配を利用した静止画像圧縮法を拡張し,3次元時空間勾配を利用した動画像圧縮へ適用したものである.実験の結果,従来手法と比較してより効率の良い時空間予測が実現できるものである.

    CiNii

  • A Study on PCA-based Fourier Descriptor in Complete and Incomplete Contour Representations

    Li TIAN, Sei-ichiro KAMATA

    Proc. of subspace2007: Subspace 2007 Workshop on ACCV2007   1 ( 1 ) 75 - 81  2007年11月

  • Automatic Image-Map Alignment by Maximization of Edge-Based Code Mutual Information

    Li TIAN, Sei-ichiro KAMATA

    Proc. of IEVC2007: IIEEJ Image Electronics and Visual Computing Workshop   35 ( 1 ) 1 - 4  2007年11月

  • Adaptive Tone Reproduction for High Dynamic Range Image

    Jian ZHANG, Sei-ichiro KAMATA

    Proc. of IEVC2007: IIEEJ Image Electronics and Visual Computing Workshop   35 ( 1 ) 1 - 4  2007年11月

  • Navigation of an autonomous sewer inspection robot based on stereo camera images and laser scanner data

    Alireza Ahrary, Li Tian, Sei-Ichiro Kamata, Masumi Ishikawa

    INTERNATIONAL JOURNAL ON ARTIFICIAL INTELLIGENCE TOOLS   16 ( 4 ) 611 - 625  2007年08月  [査読有り]

     概要を見る

    Sewer environment is composed of cylindrical pipes, in which only a few landmarks such as manholes, inlets and pipe joints are available for localization. This paper presents a method for navigation of an autonomous sewer inspection robot in a sewer pipe system based on detection of landmarks. In this method, location of an autonomous sewer inspection robot in the sewer pipe system is estimated from stereo camera images. The laser scanner data are also used to ensure accurate localization of the landmarks and reduce the error in distance estimation by image processing. The method is implemented and evaluated in a sewer pipe test field using a prototype robot, demonstrating its effectiveness.

    DOI

    Scopus

    12
    被引用数
    (Scopus)
  • Near-Duplicate Detection Using A New Framework of Constructing Accurate Affine Invariant Regions

    Li TIAN, Sei-ichiro KAMATA

    Proc. of The 9th Int. Conf. on Visual Information Systems   9 ( 1 ) 1 - 12  2007年06月

  • A Simple Tone Mapping for High Dynamic Range Image Visualization Using a Pseudo-Hilbert Scan

    Jian ZHANG, Sei-ichiro KAMATA, Li TIAN

    Proc. of IAPR Conf. on Machine Vision Applications(MVA2007)   10 ( 1 ) 363 - 366  2007年05月

  • D-12-116 点パターンマッチングための非線形最小二乗フィッティングに基づく変換パラメーター推定(D-12.パターン認識・メディア理解,一般講演)

    田 黎, 鎌田 清一郎

    電子情報通信学会総合大会講演論文集   2007 ( 2 ) 232 - 232  2007年03月

    CiNii

  • A pseudo-hilbert scan for arbitrarily-sized arrays

    Jian Zhang, Sei-Ichiro Kamata, Yoshifumi Ueshige

    IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences   E90-A ( 3 ) 682 - 690  2007年

     概要を見る

    The 2-dimensional (2-D) Hilbert curve is a one-to-one mapping between 2-D space and one-dimensional (1-D) space. It is studied actively in the area of digital image processing as a scan technique (Hilbert scan) because of its property of preserving the spacial relationship of the 2-D patterns. Currently there exist several Hilbert scan algorithms. However, these algorithms have two strict restrictions in implementation. First, recursive functions are used to generate a Hilbert curve, which makes the algorithms complex and computationally expensive. Second, both sides of the scanned rectangle must have same size and each size must be a power of two, which limits the application of the Hilbert scan greatly. In this paper, a Pseudo-Hilbert scan algorithm based on two look-up tables is proposed. The proposed method improves the Hilbert scan to be suitable for real-time processing and general application. The simulation indicates that the Pseudo-Hilbert scan can preserve point neighborhoods as much as possible and take advantage of the high correlation between neighboring lattice points. It also shows competitive performance of the Pseudo-Hilbert scan in comparison with other scan techniques.

    DOI

    Scopus

    19
    被引用数
    (Scopus)
  • A new framework for constructing accurate affine invariant regions

    Li Tian, Sei-Ichiro Kamata

    IEICE Transactions on Information and Systems   E90-D ( 11 ) 1831 - 1840  2007年

     概要を見る

    In this study, we propose a simple, yet general and powerful framework for constructing accurate affine invariant regions. In our framework, a method for extracting reliable seed points is first proposed. Then, regions which are invariant to most common affine transformations can be extracted from seed points by two new methods the Path Growing (PG) or the Thresholding Seeded Growing Region (TSGR). After that, an improved ellipse fitting method based on the Direct Least Square Fitting (DLSF) is used to fit the irregularly-shaped contours from the PG or the TSGR to obtain ellipse regions as the final invariant regions. In the experiments, our framework is first evaluated by the criterions of Mikolajczyk's evaluation framework [1], and then by near-duplicate detection problem [2]. Our framework shows its superiorities to the other detectors for different transformed images under Mikolajczyk's evaluation framework and the one with TSGR also gives satisfying results in the application to near-duplicate detection problem. Copyright © 2007 The Institute of Electronics, Information and Communication Engineers.

    DOI

    Scopus

  • A Pseudo-Hilbert scan for arbitrarily-sized cuboid region

    Jian Zhang, Sei-Ichiro Kamata

    Sixth IEEE International Symposium on Signal Processing and Information Technology, ISSPIT     764 - 769  2007年

     概要を見る

    The 3-dimensional (3-D) Hilbert scan is a one-to-one mapping between 3-D data and 1-D data along the 3-D Hilbert curve. It has been applied widely in image processing, such as image compression, object recognition, and image clustering, etc. Now, although there exist some 3-D Hilbert scanning algorithms, they usually have strict limitation on the scanned region. This makes Hilbert scan difficult to be applied in practice. So an effective scanning algorithm for arbitrarily-sized cuboid region is significant to improve the correlative digital image processing technology. In this paper, we proposed a novel Pseudo-Hilbert scanning algorithm based on the look-up tables method for arbitrarily-sized cuboid region. Although the proposed algorithm is designed for 3-D space scanning, it can be also applied in an arbitrary-sized rectangle. The algorithm does not only remove the strict constrains but also reserve the good property of the Hilbert curve preserving point neighborhoods as much as possible. The good performance of the algorithm is demonstrated by the simulation results. © 2006 IEEE.

    DOI

    Scopus

  • Near-duplicate detection using a new framework of constructing accurate affine invariant regions

    Li Tian, Sei-Ichiro Kamata

    ADVANCES IN VISUAL INFORMATION SYSTEMS   4781   61 - 72  2007年  [査読有り]

     概要を見る

    In this study, we propose a simple, yet general and powerful framework for constructing accurate affine invariant regions and use it for near-duplicate detection problem. In our framework, a method for extracting reliable seed points is first proposed. Then, regions which are invariant to most common affine transformations are extracted from seed points by a new method named the Thresholding Seeded Growing Region (TSGR). After that, an improved ellipse fitting method based on the Direct Least Square Fitting (DLSF) is used to fit the irregularly-shaped contours of TSGRs to obtain ellipse regions as the final invariant regions. At last, SIFT-PCA descriptors are computed on the obtained regions. In the experiment, our framework is evaluated by retrieving near-duplicate in an image database containing 1000 images. It gives a satisfying result of 96.8% precision at 100% recall.

  • An N-dimensional Pseudo-Hilbert scan algorithm for an arbitrarily-sized hypercuboid

    Jian Zhang, Sei-ichiro Kamata

    IECON 2007: 33RD ANNUAL CONFERENCE OF THE IEEE INDUSTRIAL ELECTRONICS SOCIETY, VOLS 1-3, CONFERENCE PROCEEDINGS   33 ( 1 ) 2459 - 2464  2007年  [査読有り]

     概要を見る

    The N-dimensional (N-D) Hilbert curve is a oneto-one mapping between N-D space and one-dimensional (I-D) space. It is studied actively in the area of digital image processing as a scan technique (Hilbert scan) because of its property of preserving the spacial relationship of the N-D patterns. Currently there exist several Hilbert scan algorithms. However, these algorithms have two strict restrictions in implementation. First, recursive functions are used to generate a Hilbert curve, which makes the algorithms complex and computation ally expensive. Second, all the sides of the scanned region must have same size and each size must be a power of two, which limits the application of the Hilbert scan greatly. In this paper, a nonrecursive N-D Pseudo-Hilbert scan algorithm based on two look-up tables is proposed. The merit of the algorithm is that the computation is fast and the implementation is much easier than the original one. The simulation indicates that the Pseudo-Hilbert scan can preserve point neighborhoods as much as possible and take advantage of the high correlation between neighboring lattice points. It also shows competitive performance of the Pseudo-Hilbert scan in comparison with other common scan techniques.

    DOI

    Scopus

  • 画像の一般変形に対する低次元の不変特徴記述子(高精細画像の処理・表示,及び一般)

    田 黎, 鎌田 清一郎

    電子情報通信学会技術研究報告. IE, 画像工学   106 ( 397 ) 41 - 44  2006年11月

     概要を見る

    本論文では、輝度の位置だけは変化し、輝度値は変化しない一般変形画像を対象とし、ヒルベルト走査に基づいた低次元の不変特徴記述子を提案する。本手法は、まずヒルベルト走査を利用して画像情報を1次元情報に変換し、その1次元情報を曲線と見なして面積を計算するものである。これは、ヒルベルト走査の性質により、曲線下の面積が一般変形に不変となる。本記述子は、従来の記述子と比較して次元が低く、一般変形に対して不変となることを確認した。

    CiNii

  • 3次元空間における一般ヒルベルト走査(高精細画像の処理・表示,及び一般)

    張 兼, 鎌田 清一郎

    電子情報通信学会技術研究報告. IE, 画像工学   106 ( 397 ) 35 - 39  2006年11月

     概要を見る

    The three-dimensional Hilbert curve is a one-to-one mapping between three-dimensional (3-D) space and one-dimensional (1-D) space. Due to the advantage of preserving high correlation of 3-D points, it receives much attention in many areas. Especially as a scan technique (Hilbert scan), the Hilbert curve is studied actively in image processing. Although there exist several Hilbert scan algorithms now, they have strict implementation conditions. First, they use recursive functions to generate scanning curves, which makes the algorithms complex and difficult to implement in real-time systems. Second, the scanned region must be a square and the length must be the power of two, which limits the application of the Hilbert scan greatly, In this paper, an effective 3-D Pseudo-Hilbert scan algorithm for an arbitrarily-sized cuboid is proposed. The proposed algorithm improves the Hilbert scan for general application. Moreover, it greatly reduces the computational complexity and saves storage memory by using two simple look-up tables instead of recursive functions. Therefore the algorithm is suitable for real-time processing. The experimental results show that the Pseudo-Hilbert scan preserves the most structures of the Hilbert scan. Although the proposed algorithm is 3-D case, it is also feasible in 2-D space. We believe this novel scan technique undoubtedly leads to many new applications in those areas which can benefit from reducing the dimensionality of the problem.

    CiNii

  • 複数の直交基底を用いた最近傍コードワードの高速探索法

    黒木 祥光, 高橋 幸太郎, 上繁 義史, 鎌田 清一郎

    映像情報メディア学会技術報告   30 ( 62 ) 29 - 34  2006年11月

    CiNii

  • I_029 拡散過程による自動画像地図照合ための共通特徴の抽出(I分野:画像認識・メディア理解)

    田 黎, 鎌田 清一郎, 恒吉 和幸

    情報科学技術フォーラム一般講演論文集   5 ( 3 ) 67 - 70  2006年08月

    CiNii

  • A gradient based predictive coding for lossless image compression

    Haijiang Tang, Sei-ichiro Kamata

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E89D ( 7 ) 2250 - 2256  2006年07月  [査読有り]

     概要を見る

    Natural, continuous tone images have a very important property of high correlation of adjacent pixels. Images which we wish to compress are usually non-stationary and can be reasonably modeled as smooth and textured areas separated by edges. This property has been successfully exploited in LOCO-I and CALIC by applying gradient based predictive coding as a major de-correlation tool. However, they only examine the horizontal and vertical gradients, and assume the local edge can only occur in these two directions. Their over-simplified assumptions hurt the robustness of the, prediction in higher complex areas. In this paper, we propose an accurate gradient selective prediction (AGSP) algorithm which is designed to perform robustly around any type of image texture. Our method measures local texture information by comparison and selection of normalized scalar representation of the gradients in four directions. An adaptive predictor is formed based on the local gradient information and immediate causal pixels. Local texture properties are also exploited in the context modeling of the prediction error. The results we obtained on a test set of several standard images are encouraging. On the average, our method achieves a compression ratio significantly better than CALIC without noticeably increasing of computational complexity.

    DOI

    Scopus

    10
    被引用数
    (Scopus)
  • ヒルベルト曲線による点照合ための新しい類似度計算法(一般セッション(5),CVのためのパターン認識・学習理論の新展開)

    田 黎, 鎌田 清一郎, 恒吉 和幸

    電子情報通信学会技術研究報告. PRMU, パターン認識・メディア理解   105 ( 674 ) 161 - 166  2006年03月

     概要を見る

    本研究では、ヒルベルト曲線を利用した点照合ための新しい類似度計算法について述べる.これは、ヒルベルト曲線を利用して、二次元の点情報を一次元点情報に変換し、一次元上で高速に類似度を計算するものである.点照合ための従来手法と比較して、計算量が少なく、雑音の影響を受けにくいことを確認した.

    CiNii

  • A study of bias correction methods for enhancing median edge detector prediction

    Haijiang Tang, Sei-Ichiro Kamata, Kazuyuki Tsuneyoshi

    2005 IEEE 7th Workshop on Multimedia Signal Processing   7 ( 1 ) 545 - 548  2006年

     概要を見る

    In this paper, we present three novel lossless compression approaches for gray-scale continuous tone natural image. Our methods enhance the median edge detector (MED), which is the core part of JPED-LS algorithm, by reducing the entropy of the prediction error via adaptive regression. These modified predictors improve the prediction accuracy by reducing the negative effect due to MED's oversimplified edge orientation detection. The experimental results show that our approaches achieve evidently better performance than MED with only neglectable increasing of computational complexity and without introduce extra pixels into the causal template.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • 指紋画像の高精度特異点抽出

    許 霄, 鎌田 清一郎, 黒木 祥光

    電気関係学会九州支部連合大会講演論文集   2006 ( 0 ) 488 - 488  2006年

    CiNii

  • 2次元空間における擬似ヒルベルト走査

    張 兼, 鎌田 清一郎, 上繁 義史

    電気関係学会九州支部連合大会講演論文集   2006 ( 0 ) 478 - 478  2006年

    CiNii

  • A fast and accurate algorithm for matching images using Hilbert scanning distance with threshold elimination function

    L Tian, SI Kamata, K Tsuneyoshi, HJ Tang

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS   E89D ( 1 ) 290 - 297  2006年01月  [査読有り]

     概要を見る

    To find the best transformation between a "model" point set and an "image" point set is the main purpose of point pattern matching. The similarity measure plays a pivotal role and is used to determine the degree of resemblance between two objects. Although some well-known Hausdorff distance measures work well for this task, they are very computationally expensive and suffer from the noise points. In this paper, we propose a novel similarity measure using the Hilbert curve named Hilbert scanning distance (HSD) to resolve the problems. This method computes the distance measure in the one-dimensional (1-D) sequence instead of in the two-dimensional (2-D) space, which greatly reduces the computational complexity. By applying a threshold elimination function, large distance values caused by noise and position errors (e.g. those that occur with feature or edge extraction) are removed. The proposed algorithm has been applied to the task of matching edge maps with noise. The experimental results show that HSD can provide sufficient information for image matching within low computational complexity. We believe this sets a new direction for the research of point pattern recognition.

    DOI

    Scopus

    20
    被引用数
    (Scopus)
  • A Pseudo-Hilbert scan algorithm for arbitrarily-sized rectangle region

    Jian Zhang, Sei-ichiro Kamata, Yoshifumi Ueshige

    ADVANCES IN MACHINE VISION, IMAGE PROCESSING, AND PATTERN ANALYSIS   4153 ( 1 ) 290 - 299  2006年  [査読有り]

     概要を見る

    The 2-dimensional Hilbert scan (HS) is a one-to-one mapping between 2-dimensional (2-D) space and one-dimensional (1-D) space along the 2-D Hilbert curve. Because Hilbert curve can preserve the spatial relationships of the patterns effectively, 2-D HS has been studied in digital image processing actively, such as compressing image data, pattern recognition, clustering an image, etc. However, the existing HS algorithms have some strict restrictions when they are implemented. For example, the most algorithms use recursive function to generate the Hilbert curve, which makes the algorithms complex and takes time to compute the one-to-one correspondence. And some even request the sides of the scanned rectangle region must be a power of two, that limits the application scope of HS greatly. Thus, in order to improve HS to be proper to real-time processing and general application, we proposed a Pseudo-Hilbert scan (PHS) based on the look-up table method for arbitrarily-sized arrays in this paper. Experimental results for both HS and. PHS indicate that the proposed generalized Hilbert scan algorithm also reserves the good property of HS that the curve preserves point neighborhoods as much as possible, and gives competitive performance in comparison with Raster scan.

  • Voting Weighted Modified Hausdorff Distance through multiscale space for automatic image-map registration

    Li Tian, Sei-ichiro Kamata

    18TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOL 4, PROCEEDINGS   18 ( 2 ) 837 - +  2006年  [査読有り]

     概要を見る

    The purpose of image-map registration is to revise the digital map included in Geographic Information System (GIS) with aerial image. The traditional method for this task requires the manual selection of tie points in both image and map. In this study, we propose a distance measure named Voting Weighted Modified Hausdorff Distance (VWMHD) for this task. In order to overcome the differences in representations between image and map in urban area, after several times of edge extraction through multiscale space, we give weights to each edge point in the initial scale based on its voting times and then compute the VWMHD for registration. The experimental results implicate that our VWMHD can provide sufficient information for automatic image-map registration and is robust to noises.

    DOI

    Scopus

    5
    被引用数
    (Scopus)
  • Diffusion geodesic path: A common feature for automatic image-map registration

    Li Tian, Sei-ichiro Kamata

    2006 IEEE INTERNATIONAL SYMPOSIUM ON SIGNAL PROCESSING AND INFORMATION TECHNOLOGY, VOLS 1 AND 2   6 ( 1 ) 944 - +  2006年  [査読有り]

     概要を見る

    The purpose of image-map registration is to revise the digital map included in Geographic Information System (GIS) with aerial image. The traditional method for this task requires the manual selection of features such as tie points in both the image and the map. Determining how to automatically extract common features between image and map is a difficult problem. In this study, we propose a novel framework of extracting the common feature named Diffusion Geodesic Path (DGP) between image and map, and use it for image-map registration. In order to overcome the differences in presentations between image and map, we first construct two new edge images from the image and the map using the diffusion process in physics to diffuse the differences between them. Then, we extract several paths between feature points (corner points) on these new edge images using the geodesic. Finally, a part of the extracted paths are automatically selected and are used for registration. The experimental results implicate that our DGP is a common and robust feature between image and map, and it can provide sufficient information for automatic image-map registration. The whole framework works well on automatic image-map registration.

  • Fast and accurate singular point extraction of fingerprint

    Xiao Xu, Sei-Ichiro Kamata

    TENCON 2006 - 2006 IEEE REGION 10 CONFERENCE, VOLS 1-4   27 ( 1 ) 1196 - +  2006年  [査読有り]

     概要を見る

    In order to maximize the robustness of fingerprint identification and to minimize the identification time, it is better to align the fingerprint first by referring the singular points of fingerprint. In this paper, we proposed a new algorithm to localize the singular points of fingerprint.
    The proposed approach is less computational complexity and higher precision. This paper shows how, from the x-gradient component of orientation filed and its sine component, very accurate extraction of singular points can be obtained. This localization can be used for accurate registration of two fingerprints.

  • A study of bias correction methods for enhancing median edge detector prediction

    Haijiang Tang, Sei-Ichiro Kamata, Kazuyuki Tsuneyoshi

    2005 IEEE 7th Workshop on Multimedia Signal Processing    2006年

     概要を見る

    In this paper, we present three novel lossless compression approaches for gray-scale continuous tone natural image. Our methods enhance the median edge detector (MED), which is the core part of JPED-LS algorithm, by reducing the entropy of the prediction error via adaptive regression. These modified predictors improve the prediction accuracy by reducing the negative effect due to MED's oversimplified edge orientation detection. The experimental results show that our approaches achieve evidently better performance than MED with only neglectable increasing of computational complexity and without introduce extra pixels into the causal template.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • A low-complexity deformation invariant descriptor

    Li Tian, Sei-ichiro Kamata

    18TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOL 2, PROCEEDINGS   2 ( 1 ) 227 - +  2006年  [査読有り]

     概要を見る

    In this paper, we propose a descriptor which is invariant to general deformations (only intensity locations change but not their value) by using Hilbert scanning. In our method, an image is converted to a 1-D sequence through Hilbert scanning at first. Then, we embed this sequence as a 1-D curve in the 2-D space. Because Hilbert scanning preserves the coherence in a 2-D image, it is easily to understand that the area under the curve is invariant to intensity location changes, naturally. Hence, we use some areas for an interest point as a deformation invariant descriptor. This descriptor can be computed in the 2-D space efficiently than other approaches where an image is embedded in the 3-D space or the dimensions of descriptors are very large. The experimental results show that our descriptor is low-complexity and superior to other approaches on interest point matching in deformation images.

    DOI

    Scopus

    2
    被引用数
    (Scopus)
  • An efficient algorithm for point matching using Hilbert Scanning Distance

    Li Tian, Sei-ichiro Kamata

    18TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOL 3, PROCEEDINGS   3 ( 2 ) 873 - +  2006年  [査読有り]

     概要を見る

    A fast and accurate similarity named Hilbert Scanning Distance(HSD) [9] has recently been presented for point matching. In this study, we improved an efficient algorithm of search strategy for HSD in the large search space. This search strategy is associated with two ideas: a relaxation greedy search, and an accelerating process using Monte Carlo sampling. The experimental results implicate that this improved algorithm is robust and efficient for point matching using HSD. It also makes a tradeoff between accuracy and speed under different requirements.

    DOI

    Scopus

  • Fast and accurate singular point extraction of fingerprint

    Xiao Xu, Sei-Ichiro Kamata

    TENCON 2006 - 2006 IEEE REGION 10 CONFERENCE, VOLS 1-4   2   1196 - +  2006年  [査読有り]

     概要を見る

    In order to maximize the robustness of fingerprint identification and to minimize the identification time, it is better to align the fingerprint first by referring the singular points of fingerprint. In this paper, we proposed a new algorithm to localize the singular points of fingerprint.
    The proposed approach is less computational complexity and higher precision. This paper shows how, from the x-gradient component of orientation filed and its sine component, very accurate extraction of singular points can be obtained. This localization can be used for accurate registration of two fingerprints.

    DOI

    Scopus

  • Diffusion geodesic path: A common feature for automatic image-map registration

    Li Tian, Sei-ichiro Kamata

    2006 IEEE INTERNATIONAL SYMPOSIUM ON SIGNAL PROCESSING AND INFORMATION TECHNOLOGY, VOLS 1 AND 2     944 - +  2006年  [査読有り]

     概要を見る

    The purpose of image-map registration is to revise the digital map included in Geographic Information System (GIS) with aerial image. The traditional method for this task requires the manual selection of features such as tie points in both the image and the map. Determining how to automatically extract common features between image and map is a difficult problem. In this study, we propose a novel framework of extracting the common feature named Diffusion Geodesic Path (DGP) between image and map, and use it for image-map registration. In order to overcome the differences in presentations between image and map, we first construct two new edge images from the image and the map using the diffusion process in physics to diffuse the differences between them. Then, we extract several paths between feature points (corner points) on these new edge images using the geodesic. Finally, a part of the extracted paths are automatically selected and are used for registration. The experimental results implicate that our DGP is a common and robust feature between image and map, and it can provide sufficient information for automatic image-map registration. The whole framework works well on automatic image-map registration.

    DOI

    Scopus

    6
    被引用数
    (Scopus)
  • H.264/AVCの直交変換におけるパディング法

    黒木 祥光, 廣重徹, 上繁 義史, 鎌田 清一郎

    情報処理学会研究報告オーディオビジュアル複合情報処理(AVM)   2005 ( 98 ) 17 - 22  2005年10月

     概要を見る

    本論文は,任意形状画像の符号化においてしばしば用いられるパディングについて考察するものである.パディングは,一般に,任意形状領域内の画素を用いて領域外部を埋めて矩形ブロックを作成することを意味し,動画像符号化MPEG4においても,low-pass extrapolation (LPE) と称する手法が採用されている.1次元DCTに対するパディング法として,既に,ShenとLiouは,領域内画素数と符号化すべきDCT係数の個数が等しいことを保証する手法を報告している.彼等は,同時に,画像信号への適用法も2種類提案しているが,彼等の手法は1次元DCTを基本としているため,領域内画素と符号化すべきDCT係数の個数が一致しない.本論文では,Shenらの手法をH.264/AVCで用いられる整数精度DCTを用いた2次元DCTに拡張する.提案する手法では,直交変換後の符号化処理を考慮し,領域内画素数と同数のDCT係数がジグザグ順序の初めに出現するため,Shenらの手法に比べ,更なる符号量の削減が期待できる.This paper presents a new padding technique for arbitrarily shaped coding. Padding background pixels using pixels in shapes, square blocks are generated; therefore, traditional orthogonal transforms work availably. A padding technique called low-pass extrapolation (LPE) is indeed employed in MPEG-4. Shen and Liou demonstrate a sophisticated padding for 1D-DCT, which guarantees that as many high frequency DCT coefficients as background pixels become zero. In order to apply their method to 2D-DCT, they also show two solutions. However, the solutions do not accomplish the equivalence between the numbers of the high frequency coefficients and the background pixels because the proposals are merely based on the 1D-DCT. In this paper, we enhance their padding methods to the orthogonal transform used in H.264/AVC, namely integer DCT. In the proposed method, the number of the DCT coefficients to be coded is equivalent to the number of the pixels lie in the shape and is appear in the zigzag order.

    CiNii

  • A new image matching algorithm for change detection using Hilbert curve

    Li Tian, Sei-ichiro Kamata, Yoshifumi Ueshige, Yoshimitsu Kuroki

    Proc. of International Astronautical Congress   56 ( B1.P.04 ) 1 - 6  2005年10月

     概要を見る

    Finding significant change in high resolution sensed image is an important task in maintaining GIS database. A class of these algorithms detects changed regions by means of edge comparisons. After extraction of feature points from a sensed image and a reference image, the feature points matching is a pivotal key in change detection. In general, given two point sets, find the minimum or maximal value of some measuring distances under the (affine) transformation. Because of the measurement errors and some outlying points, it is important that the measuring distances should be robust. Recently, a well known robust measuring distance called (partial) Hausdorff distance is widely used in feature points matching. It is more efficient than other conventional methods and has been applied in many fields. Although it is a reliable similarity measure, it is also a computational task. In this paper, we present a new algorithm using Hubert curve in order to resolve the computational complexity problem. This distance can be computed in the 1-D space rather than in the 2-D space that can greatly reduce the computational complexity. Our algorithm shows good performances for this task.

  • Image compression and retrieval using Hilbert curve

    Noritatsu Matsuo, Sei-ichiro Kamata, Kazuyuki Tsuneyoshi

    Proc. of International Astronautical Congress   56 ( B1.P.09 ) 1 - 6  2005年10月

     概要を見る

    Huge amount of high resolution satellite images for remote sensing are available in a couple of decades. Image compression method is required or highly desired in this task. Image coding standards such as JPEG and MPEG are widely used for compressing these images. However, JPEG (or MPEG) compressed data stream can not be processed without decoding. In this paper, we concentrate on image compression and retrieval without decoding by using Hilbert curve. Hubert curve is one of space filling curves which pass through all points in a space. The merits of our compression method is that (1) the compression ratio is almost equivalent to JPEG2000, (2) the computation is quite simple, (3) the required memory is rather small. This paper presents that our compression algorithm can be utilized for image retrieval from image database without decoding the compressed data stream. The block color histogram generated from the compressed data stream is used to compute the similarity measure between two images. The experimental results show that our approaches achieve better performance than conventional methods.

  • DCT係数に対するベッセル分布の適合性

    黒木 祥光, 上繁 義史, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2005 ( 0 ) 379 - 379  2005年

    CiNii

  • 走査パターンに着目したカラードキュメント画像の可逆圧縮

    古海 頌悟, 鎌田 清一郎

    電気関係学会九州支部連合大会講演論文集   2005 ( 0 ) 374 - 374  2005年

    CiNii

  • 時空間ヒルベルト走査を用いた動画像の可逆圧縮

    塚野 真司, 鎌田 清一郎, 上繁 義史, 黒木 祥光

    電気関係学会九州支部連合大会講演論文集   2005 ( 0 ) 373 - 373  2005年

    CiNii

  • 画像照合のためのヒルベルト走査距離

    田 黎, 鎌田 清一郎, 恒吉 和幸

    電気関係学会九州支部連合大会講演論文集   2005 ( 0 ) 544 - 544  2005年

    CiNii

  • An automatic image-map registration algorithm using modified partial Hausdorff distance

    L Tian, SI Kamata, Y Ueshige, Y Kuroki

    IGARSS 2005: IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, VOLS 1-8, PROCEEDINGS   25 ( 6 ) 3534 - 3537  2005年  [査読有り]

     概要を見る

    The purpose of image-map registration is to revise the digital map included in Geographic Information System (GIS) with Remote Sensing Images (RSI). Because edge points extracted from RSI are very different from those extracted from digital maps, the error of registration different edge points propagates to the final registration results. In this paper, we provide an automatic algorithm for the image-map registration problem. A novel concept of modified Partial Hausdorff distance (MPHD) is proposed its :I distance measure. This measure is quite tolerant of the differences between map and image. In order to decrease the computational complexity, we also use Hilbert curve to select the initial candidates of edge points in our algorithm. The number of initial candidates can be adjusted to meet the required search accuracy and/or speed. The experimental results implicate that MPHD of our algorithm could provide sufficient information for image-map registration.

  • A cooperative stereo matching algorithm for sewer inspection robots

    A Ahrary, L Tian, S Kamata, M Ishikawa

    PROCEEDINGS OF THE SIXTH IASTED INTERNATIONAL CONFERENCE ON ROBOTICS AND APPLICATIONS   11 ( 1 ) 294 - 299  2005年  [査読有り]

     概要を見る

    Stereo matching is an essential issue in computer vision. Recently, many stereo matching algorithms based on segmentation, graph cuts and so on have been proposed. Because the disparities change continuously in sewer environment, these methods are not applicable to sewer systems and are computationally expensive. In this paper, we propose a new cooperative stereo matching algorithm that is suitable for this task. In our algorithm, a reference image is divided into a feature pixel group (edges, cracks, etc.) and a non-feature pixel group (walls, etc.). Then, a proposed matching measure named Linear Computation is implemented in the feature pixel group. Also, the conventional measure of the Sum of Squared Differences (SSD) is implemented in the non-feature pixel group. In order to improve the accuracy of the conventional SSD measure, we also impose constraints such as an evidence constraint and a neighboring similarity constraint on the SSD measure. The proposed cooperative algorithm achieves better results than other conventional algorithms in the sewer environment.

  • Matching vehicles using Hilbert scanning distance

    L Tian, S Kamata, K Tsuneyoshi

    2005 IEEE INTERNATIONAL CONFERENCE ON VEHICULAR ELECTRONICS AND SAFETY PROCEEDINGS   2005 ( 1 ) 149 - 154  2005年  [査読有り]

     概要を見る

    Matching objects is a fundamental problem for any object detection system. Feature-based methods in matching objects such as vehicles often encounter the problem of correspondences between features of two related patterns. The features may be points, fines, curves and regions. Point pattern matching (PPM) is a primary and essential approach for establishing a correspondence within two related patterns. Although some well-known Hausdorff distance measures work well for this task, they are very computational expensive and suffer from the noise of images. In this paper, we propose a novel similarity measure using Hilbert curve named Hilbert scanning distance (HSD) to resolve the problems. This method computes the distance measure in one-dimensional (I-D) sequence in stead of in two-dimensional (2-D) image space, which greatly reduce the computational complexity. By applying a threshold elimination function, extreme distances caused by noise and position errors (e.g. those that occur with feature or edge extraction) are removed. The experimental results show that HSD can provide sufficient information for matching vehicles within low computational complexity. We believe this point out a new direction for the research of PPM.

    DOI

    Scopus

    1
    被引用数
    (Scopus)
  • An autonomous sewer robots navigation based on stereo camera information

    A Ahrary, L Tian, S Kamata, M Ishikawa

    ICTAI 2005: 17TH IEEE INTERNATIONAL CONFERENCE ON TOOLS WITH ARTIFICIAL INTELLIGENCE, PROCEEDINGS   2005 ( 1 ) 628 - 633  2005年  [査読有り]

     概要を見る

    In this paper we propose a method for an autonomous sewer robots to navigate through a sewer pipe system based on stereo camera information. In this method, a local features such as manholes and pipe joints are extracting as a feature pixels in the Region of Interest (ROI) of left image. Then, an accurate and fast stereo matching measure named Linear Computation is implemented in this ROI image to compute the distance between the robots and local features. Finally, the distance data can be used for navigation map in sewer pipe system. The experimental results show that our method can provide sufficient information for an autonomous sewer robots navigation.

    DOI

    Scopus

    15
    被引用数
    (Scopus)
  • 空間充填曲線と画像処理応用

    鎌田 清一郎

    情報処理学会研究報告グラフィクスとCAD(CG)   2004 ( 121 ) 25 - 30  2004年11月

     概要を見る

    G.ペアノ(Peano)は 1890年『平面領域内の全ての点を通過するような曲線』を発見し その存在を明らかにした. 現在 線分を単位超立方体全体へ移すこのような連続曲線は 空間充填曲線 あるいはペアノ曲線と呼ばれている.空間充填曲線の中で 応用研究の最も多い曲線はヒルベルト曲線である.例えば ヒルベルト曲線の応用としては画像圧縮 スペクトル画像分類 データベース情報検索 計算機ホログラムなど 様々な分野に及ぶ.本論文では 空間充填曲線について定義と3つの例を紹介し 次にヒルベルト曲線を中心とした画像処理への応用研究を幾つか概観する.In 1890, G.Peano found a curve which passes through all points in a space and proved its existance. So far such a countinuous curve which maps a unit interval into a unit hypercube is called a space-filling curve (SFC) or a Peano curve. There are several applications in the area of image processing, computer graphics, database retrieval, etc. Among the SFC's, the most applicable curve is a Hilbert curve. This paper describes the definition and some examples of the SFC, and then overviews image processing applications of the SFC, especially the Hilbert curve.

    CiNii

  • 静止画像の動的予測符号化における演算負荷の少ない予測器の選定法

    黒木 祥光, 上繁 義史, 鎌田 清一郎

    情報処理学会研究報告オーディオビジュアル複合情報処理(AVM)   2004 ( 99 ) 13 - 18  2004年10月

     概要を見る

    ディジタル画像データを無歪みで圧縮する手法として,一般に予測符号化が用いられる.予測符号化は符号化済みの画素を用いて符号化すべき画素を予測し,予測誤差をエントロピ符号化する手法である.線形予測における最適な予測係数は,予測誤差電力を最小化するという観点から,正規方程式により算出される.正規方程式の解は,実数で与えられるが,演算負荷の少ない予測器を実現するには,数回のシフト演算と加減算で算出可能な値,例えば±1/2,±1/4,±3/4といった値が好ましい.本研究では,線形予測に用いる画素を,JPEG,JPEG-LSの可逆モードと同様に,着目画素の近傍3画素とし,予測係数の代数和は1であるとの条件の下,数回のシフト演算と加減算のみで算出可能な,演算負荷の少ない予測器に対する予測誤差電力の定量的評価を通じ,各予測器が最適となる条件を示す.また,11種および6種類から成る予測器の組を提案し,ブロック適応予測に使用した場合の性能評価も示す.Predictive coding is generally employed in lossless coding algorithms for digitized images. Pixel values to be coded in the predictive coding schemes are predicted from already-coded adjacent pixels, and then, an entropy coder encodes the resulting prediction residuals. This paper describes an estimate and a selection method of the simple predictors in a theoretical viewpoint, which constrained as follows: (1) The predicted values are computed by weighted summation of the neighboring three pixels identical to JPEG and JPEG-LS; (2) The predicted values are obtained with a small number of shift operations, additions, and/or subtractions; (3) The predictor coefficients sum up to 1. The comparison of the total squared error between the simple predictors and the optimum predictors obtained by solving the normal equations is discussed. With the results of the theoretical analysis, we propose a six-predictor and an eleven-predictor set predictors. The performance of the proposed predictors on actual images is also presented.

    CiNii

  • 複数走査を用いた自然画像の可逆圧縮法(画像・映像処理)

    小林 正明, 鎌田 清一郎

    電子情報通信学会論文誌. D-II, 情報・システム, II-パターン処理   87 ( 8 ) 1603 - 1612  2004年08月

     概要を見る

    画像通信,蓄積においては膨大な情報量を処理するために画像圧縮が求められる.医療分野や歴史的文化財などのディジタルアーカイビング分野で使用される自然画像は,データの精度が重要であり,可逆圧縮が要求される.白熱画像に対する圧縮方法は,JPEG-LSの標準化作業においてもいくつかの方法が提案され,その多くが予測符号化に基づいている.自然画像は2次元方向での近傍画素間に高い相関があり,また,エッジ,テクスチャ,グラデーションなどによって局所的に類似した濃度変化をする.しかし,従来手法はラスタ走査順に符号化を行っているため,十分にこれらの冗長度が削減されていないと考えられる.そこで,本論文では,正方領域単位で画像を分割し,ラスタ走査順ではなく複数の走査パターンから最適な走査を選択し,選択された走査に沿って予測式の重み係数を適応的に更新する予測符号化について検討を行った.いくつかの自然画像を用いた評価実験の結果,JPEG-LSに比べ2〜10%程度高い符号化効率が実現できることを確認した.

    CiNii

  • Efficient Updating Digital Maps by Remote Sensing Images and Their Usage for Applied Tasks

    Sergey Ablameyko, Aleksandr Kryuchkov, S.Borichev, Sei-ichiro Kamata

    Proc. of 24th Int. Sympo. on Space Technology and Science   24 ( 1 ) 1 - 5  2004年06月

  • 北九州学術研究都市 早稲田大学大学院情報生産システム研究科メディア情報部門

    岡田 稔, 鎌田 清一郎

    画像電子学会誌 = Imaging & Visual Computing The Journal of the Institute of Image Electronics Engineers of Japan   33 ( 2 ) 289 - 293  2004年03月

    CiNii

  • A modified method of adaptive space-filling coding

    Yoshifumi Ueshige, Sei-ichiro Hiratsuka, Sei-ichiro Kamata

    Proc. of Int. Workshop on Advanced Image Technology   7 ( 1 ) 177 - 180  2004年01月

  • A Method of Computing a Space Filling Curve for Arbitrarily Shaped Region

    Sei-ichiro Hiratsuka, Yoshifumi Ueshige, Sei-ichiro Kamata

    Proc. of Int. Workshop on Advanced Image Technology   7 ( 1 ) 181 - 184  2004年01月

  • A multi-scan adaptive linear prediction approach for lossless image compression

    HJ Tang, SI Kamara, MA Kobayashi

    2004 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN & CYBERNETICS, VOLS 1-7   34 ( 5 ) 3004 - 3009  2004年  [査読有り]

     概要を見る

    This paper presents a block based lossless compression approach for gray scale images. Multiple scanning methods are applied to each block, and a newly proposed adaptive linear prediction is performed There are different prediction residuals obtained corresponding to different context based on multiple scanning. We choose the best residual for coding. That is, rather than relying any single scanning, our approach is to select a scanning produces the best result on each black. The prediction coefficients are updated during the scanning to optimize the coding accuracy. Experiment results show that our method out-performed JPEG-LS 4similar to5% in compression efficiency.

  • Lossless image compression via multi-scanning and adaptive linear prediction

    HJ Tang, S Kamata, K Tsuenyoshi, M Kobayashi

    PROCEEDINGS OF THE 2004 IEEE ASIA-PACIFIC CONFERENCE ON CIRCUITS AND SYSTEMS, VOL 1 AND 2   7 ( 1 ) 81 - 84  2004年  [査読有り]

     概要を見る

    This paper presents an efficient lossless compression approach for gray scale images. The main contributions are 1) we divide the input image to blocks of pixels and use different scanning methods, we select the one that minimizes the prediction error for coding; 2) a new adaptive linear prediction is applied. Experiment results show that our method outperformed JPEG-LS 2 similar to 10% in compression efficiency.

  • 複数走査による可逆画像圧縮法

    唐 海生, 鎌田 清一郎, 小林 正明

    映像情報メディア学会技術報告   27 ( 72 ) 77 - 82  2003年12月

    CiNii

  • A new probability density function of DCT coefficients

    Yoshimitsu Kuroki, Yoshifumi Ueshige, Sei-ichiro Kamata

    Proc. of 3rd Int. Workshop on Spectral Methods and Multirate Signal Processing   3 ( 1 ) 129 - 133  2003年09月

  • カラードキュメント画像の可逆圧縮法

    小林 正明, 鎌田清一郎

    電子情報通信学会論文誌(D-II)   J85-D-II ( 4 ) 584 - 593  2002年04月

     概要を見る

    プリンティングシステムにおいては,文字,グラフィックス,自然画像等を含むカラードキュメント画像の使用頻度が高い.このため,ドキュメント画像のデータ量削減に関する要求は大きい.従来のカラードキュメント画像に対する圧縮方法としては,領域分割を行い,各領域に適した圧縮方法を適用する方法が検討されている.しかし,領域分割に時間がかかること,誤判定により符号化効率や画質が低下すること,復号時の画像合成処理に多くの記憶容量を要することなど,いくつかの課題が残されている.そこで,本論文では領域分割を必要としないカラードキュメント画像の可逆圧縮方法について検討を行った.本手法は,バンド単位で走査変換を行うことにより近傍保存性の高い1次元データを作成し,作成された1次元データに対してランレングス符号化を行うものである.ランレングス符号化においては近傍に出現するランの色及びランの長さに非常に強い相関があるという特徴を利用して符号化効率の向上を図った.いくつかのカラードキュメント画像を用いた実験結果からGzipに比べ19%,JPEG-LSに比べ50%高い符号化効率が実現できることを確認した.

    CiNii

  • カラー静止画像の高速可逆圧縮方法

    小林 正明, 鎌田 清一郎

    画像電子学会誌   31 ( 5 ) 778 - 786  2002年

     概要を見る

    RGBカラー静止画像をR-,G-,B-の三つの色プレーンに分けた場合,各色プレーンごとの画像は高い相関を持つことが知られている.また,各画像は局所領域ごとに異なる性質(コンテクスト)を持つことも知られている.本論文では,これらの性質を利用することにより画像の持つ冗長度を除去しRGBカラー静止画像の可逆圧縮を行う方法について提案する.色プレーンごとに予測変換を行った予測誤差データに対して色プレーン間の相関を利用して予測誤差の色差成分を生成し,生成された予測誤差の色差成分を局所的な性質を利用してコンテクストごとに分離し,分離されたコンテクストごとに符号化を行う.従来手法との比較実験から,LOCO-I, CALICに比べ符号化効率が14%,13%程度改善され,CREWに対して同等以上の符号化効率を実現できることを確認した.また,処理時間はこれらの従来手法より高速に実現できることを確認した.

    DOI CiNii

    Scopus

  • Lossless Compression for Compound Color Document Images

    Masa-aki Kobayashi, Sei-ichiro Hiratsuka, Sei-ichiro Kamata

    Proc. of the 21st Int. Display Research Conf.   21 ( 2 ) 1525 - 1528  2001年10月

  • Color thinning with applications to biomedical images

    Alexandr Nedzved, Yurii Ilyich, Sergey Ablameyko, Sei-ichiro Kamata

    Proc. of 9th Int. Conf. Computer Analysis of Images and Patterns   9 ( 1 ) 256 - 263  2001年09月

  • N-dimensional Hilbert scanned hierachical histogram representation for cluster analysis

    Sei-ichiro Kamata, Sergey Ablameyko

    Proc. of 6th Int. Conf. on Pattern Recognition and Information Processing   6 ( 1 ) 113 - 120  2001年05月

  • Correlation binary image processing based on matrix representation

    R.Bogush, S.Maltsev, Sergey Ablameyko, Sei-ichiro Kamata

    Proc. of 6th Int. Conf. on Pattern Recognition and Information Processing   6 ( 1 ) 87 - 94  2001年05月

  • 時空間ヒルベルト走査を用いた適応的線形予測符号化による動画像可逆圧縮

    坂東 幸浩, 横山 貴政, 鎌田 清一郎

    映像情報メディア学会誌   55 ( 3 ) 447 - 454  2001年03月

     概要を見る

    In video processing, compression techniques are required to improve the efficiency of transmission and storage of video data. Data compression methods can be classified into two basically different categories : lossy compression methods and lossless compression methods. Lossless compression methods are required for medical image processing and for compressions broadcast materials. We developed a novel video lossless compression method using the Hilbert curve, by mapping the three-dimensional data to one-dimensional data along the curve and then applying adaptive linear prediction coding to the one-dimensional data. Experimental results show that better compression ratio is obtained with our method than with lossless JPEG, although the processing time for both methods is almost equal.

    DOI CiNii

    Scopus

  • Region-based image coding with multiple algorithms

    M Petrou, PX Hou, S Kamata, CI Underwood

    IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING   39 ( 3 ) 562 - 570  2001年03月  [査読有り]

     概要を見る

    The wide usage of small satellite imagery, especially its commercialization, makes data-based onboard compression not only meaningful but also necessary in order to solve the bottleneck between the huge volume of data generated onboard and the very limited downlink bandwidth. In this paper, we propose a method that encodes different regions with different algorithms. We use three shape-adaptive image compression algorithms as the candidates. The first one is a JPEG-based algorithm, the second one is based on the object-based wavelet transform (OWT) method proposed by [1], and the third adopts Hilbert scanning of the regions of interest followed by one-dimensional (1-D) wavelet transform. The three algorithms are also applied to the full image so that we can compare their performance on a whole rectangular image. We use eight Landsat Th I multispectral images and another 12 small satellite single-band images as our data set. The results show that these compression algorithms have significantly different performance for different regions. For relatively smooth regions, e.g., regions that consist of a single type of vegetation or water areas etc, the 1-D wavelet method is the best. For highly textured regions, e.g., urban areas, mountain areas, and so on, the modified OWT method wins over the others. For the whole image, OWT working at whole image mode, which is just an ordinary 2-D wavelet compression, is the heat. Based on this, we propose a new data-based compression architecture that extracts particular regions according to the application of interest and then involves different algorithms to encode different regions in order to achieve better performance than traditional onboard compression schemes in which a fixed compression method is applied to the whole image no matter what the application is. This approach is most appropriate for use with images captured by microsatellites, which are commissioned for specific applications in which one knows a priori which class of region the user is interested in.

  • An efficient correlation computation method for binary images based on matrix factorisation

    R Bogush, S Maltsev, S Ablameyko, S Uchida, S Kamata

    SIXTH INTERNATIONAL CONFERENCE ON DOCUMENT ANALYSIS AND RECOGNITION, PROCEEDINGS   6 ( 1 ) 312 - 316  2001年  [査読有り]

     概要を見る

    A novel algorithm for complexity reduction in binary image processing, namely for computation of correlation between image and object template is proposed. This algorithm is based on direct computation of vector-matrix multiplication with utilisation of binary matrix factorisation approach. Comparison with other algorithms is given and it is shown that our approach allows to reduce tithe and complexity of this task.

  • N次元空間における一般化ヒルベルト走査の一計算法

    坂東 幸浩, 鎌田清一郎

    電子情報通信学会論文誌(A)   J83-A ( 12 ) 1368 - 1381  2000年12月

     概要を見る

    空間充てん曲線の一例であるヒルベルト曲線はこの曲線のもつ自己相似性及び近傍保存性の良さから, 近年, 様々な研究に応用されている.特に, この曲線を利用した多次元データ系列から1次元データへの展開はヒルベルト走査と呼ばれている.しかし, ヒルベルト走査は1辺を2のべき乗とする超立方体領域にしか適用できないという問題がある.この走査領域の問題に対し, 本論文では, ヒルベルト走査を超直方体領域に対応可能な走査法へと拡張する.まず, 走査が可能な超直方体領域の条件を明らかにし, 次に, 走査アドレスの計算に用いる参照用のテーブルの構成法を示し, 最後に, この参照用のデーブルを用いた走査アルゴリズムを示す.

    CiNii

  • An address generator for an N-dimensional pseudo-Hilbert scan in a hyper-rectangular parallelepiped region

    Y Bandoh, S Kamata

    2000 INTERNATIONAL CONFERENCE ON IMAGE PROCESSING, VOL I, PROCEEDINGS   7 ( 1 ) 737 - 740  2000年  [査読有り]

     概要を見る

    Hilbert curve is a one-to-one mapping between N- dimensional (N-D) space and 1-D space. The Hilbert curve has been applied to image processing as a scanning technique (Hilbert Scan). Recently applications to multi-dimensional image processing are also studied. In this application, we use N-D Hilbert scan which maps N-D data to 1-D data along N-D Hilbert curve. However, N-D Hilbert scan is the application limited to data. in a hyper-cube region. In this paper, we present a novel algorithm for generating N-D pseudo-Hilbert curves in a hyper-rectangular parallelepiped region. Our algorithm is suitable for real-time processing and easy to implement, in hardware, since it is a simple and non-recursive. computation using look-up tables.

  • Region-based scanning for image compression

    S Kamata, Y Hayashi

    2000 INTERNATIONAL CONFERENCE ON IMAGE PROCESSING, VOL II, PROCEEDINGS   7 ( 1 ) 895 - 898  2000年  [査読有り]

     概要を見る

    G. Peano has published a paper of space-filling curve in 1890. There are several applications using this curve in the area of image processing, computer graphics, etc. This paper presents a region-based scanning technique for image compression which we call an adaptive space filling scan (ASFS). In order to generate the ASFS, we make use of a minimum spanning tree technique. From several experiments for image compression, it is confirmed that in comparison to JPEG, acceptable quality images can be obtained at the same bit-rates.

  • ヒトベルト走査を用いた時空間領域分割による高速動画像圧縮法

    坂東 幸浩, 西 修功, 鎌田 清一郎

    映像情報メディア学会誌 : 映像情報メディア = The journal of the Institute of Image Information and Television Engineers   53 ( 4 ) 559 - 564  1999年04月

     概要を見る

    We investigate a novel video compression method using the Hilbert curve. This curve is applied to various problems such as image compression, the traveling salesman problem, and database management, because of its locality preservation property. We map the three-dimensional data to one-dimensional data along the Hilbert curve and then apply a lossy compression method using a linear approximation to the one-dimensional data. This method uses simple segmentation of the one-dimensional data and does not require complex computation such as DCT or motion estimation. Experimental results show that our method obtains acceptable quality of reconstructed images comparable to H.263 at low bit rates but is about ten times faster than H.263 (full search motion estimation). It is suitable for video telephones which require real-time encoding of video signals.

    DOI CiNii

    Scopus

▼全件表示

書籍等出版物

  • Image Processing - Dealing with Textures -

    Maria Petrou, Sei-ichiro Kamata( 担当: 共著,  担当範囲: Second Edition)

    Wiley  2021年01月

  • 画像処理−画像表現・圧縮・フラクタル−

    鎌田清一郎

    サイエンス社  2003年03月 ISBN: 4781910297

共同研究・競争的資金等の研究課題

  • スパース・ハイパーグラフネットワークによる画像認識および検索と調剤過誤防止応用

    日本学術振興会  科学研究費助成事業

    研究期間:

    2021年04月
    -
    2024年03月
     

    鎌田 清一郎

     概要を見る

    前年度に引き続き、グラフ構造そのものを一般化したハイパーグラフニューラルネットワーク(HGNN)を対象として、スパースグラフ表現によるHGNN(SHGNN)を検討している。これは、グラフ表現へ変換する際にハイパーエッジに基づくグラフ分割を行い、サブグラフのスパース構造化そして全体をスパース最適化することで、SHGNNの構築を図ったものである。今年度は、このモデルをさらに改善するため、マルコフ確率場(MRF)を導入したMRFハイパーグラフTransformerを構築した。条件としては、ノードやハイパーエッジ間でマルコフ性を導入し、すべてのハイパーエッジをクリークとして扱い、異なるハイパーエッジ間のノードはガウス分布に従うものとし、最適なハイパーグラフ構造に再構築する方法を検討した。視覚的質問応答応用としてデータセットVQAーv2に対してMRFハイパーグラフTransformerを適用した結果、ハイパーグラフに基づく従来手法を越える適合性能を得ることができた。次に、SHGNNの応用として人物の行動認識を取り上げ、人物スケルトンに対して、スパース・ハイパーグラフの構築を行った。各関節部位をノードに割り当て、スパース表現によりどの連結が重要かを抽出し、人物スケルトンのハイパーグラフをスパース化するものである。公開されている行動認識データセットNTU-RGB+D等を用いた評価実験を行い、従来手法のHyper-GNNやDHGCNなどに比べて高い識別性能を達成した。さらに、これまで緊急の社会問題となっている薬学リスクマネジメントにおける調剤過誤防止に引き続き取り組んでいる。すなわち事故に繋がらないヒアリハットの2009年分から蓄積した約30万件の事例データセットを構築し、ヒアリハット検索を行うための用法用量などの情報抽出を継続して行っている。

  • スパースグラフ・ニューラルネットワークによる画像認識および応用

    日本学術振興会  科学研究費助成事業

    研究期間:

    2018年04月
    -
    2021年03月
     

    鎌田 清一郎

     概要を見る

    近年、ニューラルネットワーク(NN)の一般化としてグラフNN(GNN)が活発に研究されている。本研究では、スパースグラフ表現によるGNN(SGNN)について検討している。グラフのスパース性は古くから検討されてきたが、GNNに向いたスパースグラフをどのように構築すればよいか、どのようなスパース拘束条件が必要か、などいくつかの課題がある。これらを解決するために、スパース拘束条件として、(1)グラフの頂点数に関する条件、(2)グラフのエッジ数に関する拘束条件、(3)結合性に関する条件などを含んだ拘束条件を導入し、相互k-NN(Nearest Neighbor)と組み合わせたk3スパースグラフを提案した。これをベースとして新たなLoss評価関数およびPooling法を用いた、Siameseネットワークを導入したSGNN(k3SGNN)を考案し、顔画像の識別に適用した。標準顔画像データセットLFW(Labeled Faces in the Wild)などを使用した比較評価実験では、本k3SGNNがGoogleによる超多層のFaceNetとほぼ同等の認識精度を示した。次に、応用研究として、眼底画像の重症度識別による糖尿病網膜症の早期発見を行うため、眼底画像に対してスパースグラフを構築し、上述のSGNNを適用した。昨年行われた国際コンテスト「Diabetic Retinopathy: Segmentation and Grading Challenge」において、そのデータセットが公開されたので、そのデータセットを利用し、当該コンテストで第1位の方式との性能比較評価を行ったところ、提案手法の識別精度が数%向上したことを確認した。また薬学リスクマネージメントにおける調剤過誤防止実現のため、薬剤画像の識別問題に取り組んでおり、現在ヒューマンエラーによるヒアリハット発生状況を調査している。IEICE Transactions on Information and Systemsにおいて、提案手法のSGNNと顔認識応用に関する研究論文を発表した。また、英国ダラム大学ブレッコン・トビー教授と共著で、米国電気電子学会の主要国際会議であるInternational Conference on Image Processing 2018 (画像処理に関する国際会議2018)において研究論文を発表した。また、国際会議ICARCV2019では、k3スパース性を考察し、k3SGNNについて研究発表した。以上のことから、本研究はおおむね順調に進んでいる。2018年度に引き続き、スパースグラフ表現のスパース性を考察し、さらにSGNNの識別精度の向上を図りたい。またスパースグラフ表現とグラフニューラルネットワークを利用した応用研究が増えようとしており、今後は、他応用分野へのその可能性も追求していく予定である

  • 多次元画像のスパースフーリエ変換と深層学習の高速化

    日本学術振興会  科学研究費助成事業

    研究期間:

    2016年04月
    -
    2018年03月
     

    杉本 憲治郎, 鎌田 清一郎, 福嶋 慶繁, 京地 清介, 黒木 祥光, 平川 恵悟

     概要を見る

    工学において重要な高速フーリエ変換の発展形であるスパースフーリエ変換の高度化に取り組んだ.またそれに関連してスペクトルスパース性に基づく定数時間フィルタを提案し,計算量と近似精度の両面からの性能向上を実現した.当該研究期間である2016-2017年度での研究業績としての成果は,ジャーナル論文1件,学会発表16件(内訳は国際会議7件,国内会議9件),受賞3件であった.国際会議発表の多くは当該分野で最も権威あるフラグシップ会議(ICIPとICASSP)に採択され,また国内発表でも3件が受賞につながるなど,国内外で高い評価を得たと考えている

  • ビジュアルビッグデータの高速画像検索・認識に関する研究

    日本学術振興会  科学研究費助成事業

    研究期間:

    2015年04月
    -
    2018年03月
     

    鎌田 清一郎, 杉本 憲治郎, 張 ケツ石, 柳 済群, 郝 鹏翼, 呉 仁傑, 奥谷 遼, 笑夕, 頼 咏文, 林 雪コウ, 邱 帆, 湯 凱華, ゴ 欣卉, 倪 守誠, 矢野 光一, 田 黎, 馬 利庄, ブレッコン トビー

     概要を見る

    ビジュアルメディアのビッグデータ(以下、ビジュアルビッグデータとする)を利活用し、画像の検索および認識に関する研究において、できるだけ計算量を削減するための情報理論における圧縮可能性について再検討し、高精度かつ高速な画像検索および認識の方法論を確立した。特に近年注目をされているディープラーニングとの融合により、顔のビジュアルビッグデータを基にしてスパースグラフニューラルネットワークという新たな研究領域を構築することができた

  • 空間充填曲線による画像認識・検索に関する研究

    日本学術振興会  科学研究費助成事業

    研究期間:

    2012年04月
    -
    2015年03月
     

    鎌田 清一郎, 杉本 憲治郎, 張 ケツ石, 柳 済群, 楊 卓, 郝 鵬翼, 周 尉, 兪 霁野, 汪 セイ傑, 呉 仁傑

     概要を見る

    薬局等で患者に間違いなく薬が提供されているかをチェックする大規模画像検査システムを実現するため、空間充填曲線を利用して画像認識・検索に関する研究を行った。まず画像の色情報を利用した線形多様体色記述子を提案し、色特徴記述について新たな方法論を確立した。また薬剤などの形状記述には回転不変で識別能力の高い新たな多元極フーリエ記述子を提案した。さらに複数の空間充填曲線を利用して適応的ヒルベルト走査型Bag-of-Features検索方式を提案した。これは、画像の性質を基にしてヒルベルト走査などの様々な走査方法を適応的に選択する方式を導入したものであり、画像の検索効率を向上させることができた

  • 空間充填曲線による画像検索および暗号に関する研究

    日本学術振興会  科学研究費助成事業

    研究期間:

    2009年
    -
    2011年
     

    鎌田 清一郎

     概要を見る

    本研究では、MPEG7などに対して、空間充填曲線を利用した高効率な画像検索方式および暗号方式の実現を行った。大規模画像データベースの検索において、ヒルベルト走査型Bag-of-Features構造表現による高速検索手法を新たに提案した。また、高効率な暗号方式では、多次元空間を充填するヒルベルト曲線生成を利用した高速鍵暗号方式を開発した

  • 高速パターンマッチング応用

    研究期間:

    2007年
    -
    2011年
     

  • インターネット上でのプライバシ保護が可能なオンライン生体認証システムの構築

    日本学術振興会  科学研究費助成事業

    研究期間:

    2006年
    -
    2007年
     

    上繁 義史, 櫻井 幸一, 鎌田 清一郎

     概要を見る

    現在,主として生体認証はクローズ環境での認証に利用されているが,インターネットバンキングなど,将来的にオープン環境での利用が期待されている.そのような生体認証について,近年システム要件,データ形式の研究や国際標準化が進んでいるが,テンプレートのプライバシ保護技術について,十分な基礎研究がなされていない.認証とデータベース(DB)の視点から研究を行った.ワンタイム生体認証として,認証セッションごとに異なる変換処理をテンプレートと検証用の特徴情報に施すことによってワンタイム性を持たせる手法について提案した.本手法では変換生成のTTPの秘密情報,認証セッション情報等のタイムスタンプ等の情報を利用して変換関数を生成,適用することによりワンタイム性を確保する.この手法では通信路上の盗聴に対して安全性を確保することが出来る.テンプレートDBの保護方式1として,DB内部での格納アドレスを攪拌することによって,登録者ごとにテンプレートを分散格納する方法を提案した.この手法では,隣接するアドレスに同一人物のテンプレート要素が格納されることがないため,マップを入手されなければ,登録者のテンプレート全体を再構成することはできない.それゆえに成りすましを防止する効果もある.テンプレートDBの保護方式2として,テンプレートが,登録者間で相互相関をもつことを利用して,登録者間の平均値情報との差分情報を登録テンプレートとして,ストレージに格納する,テンプレートDBを提案した.提案法では,成りすまし,改ざんを困難にする効果が見込まれる.本補助金に基づく研究は,全研究過程の最初の段階に属している.平成19年度後半には,研究の発展段階として,装置への実装を行うため,アルゴリズムの開発に着手した.本成果を平成20年度に発表すべく,改良・実験を通じて準備を進めている

  • 物理モデル駆動によるノンフォトリアリスティック画像創成と知的符号化

    日本学術振興会  科学研究費助成事業

    研究期間:

    2005年
    -
    2007年
     

    岡田 稔, 鎌田 清一郎, 水野 慎士

     概要を見る

    ノンフォトリアリスティックコンピュータグラフィックス技術(NPR : Non-photorealistic rendering)は、様々な可視化技術などに使用されるが、近年では、美術・工芸の分野への応用が模索されている。本課題はこれらのNPR画像生成を、物理学的根拠のある生成方式によって行うとともに、既存の版画作品を高次レベルで符号化することを目的としている。まず、PBR(physics-based rendering)アプローチに基づき、木彫刻、木版画、銅版画を題材としたコンピュータグラフィックス(CG)画像合成法を開発した。特に木版画ではユーザインタフェースの開発を進めた。また銅版画ではドライポイントと呼ばれる芸術性が高い技術の工程と現象を物理的に模擬し、画像合成を試みた。本研究は九州産業大学芸術学部の協力を得て、専門家の参加により評価が行われたが良好な評価結果を得た。次に、次世代カーナビゲーションシステムの構成法と要素技術の開発を行った。従来型の二次記憶ベースのフルCGによるカーナビに対して、本研究では拡張現実を利用した新方式を提案した。そこでは車載カメラから得た走行前方画像をコンピュータビジョン技術により符号化し、仮想世界を構築した。その仮想世界の同一視点の画像を生成し、運転者に提示することによってリアルタイム情報を効率的に伝える方式を実現した。さらに、ノンフォトリアリスティック画像(non-photorealistic image:非写実的画像)の生成技法として、銅版画、特にメゾチントに焦点を当て、物理モデルの構成と画像生成法の研究を実施した。そこでは二種類の中間表現画像により、多彩な濃淡階調表現能力を有するメゾチント技法の再現を試みた。その結果、物理的にも良好なメゾチント作品をPBRの枠内で画像生成できることを確認した

  • ハイブリッド画像圧縮システム

    研究期間:

    2003年
    -
    2007年
     

  • 乳腺用3次元超音波画像取得装置に関する基礎研究

    日本学術振興会  科学研究費助成事業

    研究期間:

    2005年
    -
    2006年
     

    西村 敏博, 岡田 稔, 鎌田 清一郎, 椿井 正義

     概要を見る

    少子化社会を迎えようとする21世紀の壁頭において、乳がんによる原因で母親となるべき女性が若年で死亡する例が後を絶たない。本研究では乳腺組織の診断を目的として、超音波診断装置における高精細な3次元画像取得法を研究する。画像を取得する検者の技量に依存せず、誰もが簡単な操作で高精細な画像を取得できる機構の実現を目的とする。研究課題は、超音波画像の画像化における画質の判定と自動調整である。厚生労働省はX線を用いたマンモグラフィーによる乳房診断を推奨している。しかし、X線による被曝は検診の回数が重なるほど人体に対して危険である。一方、超音波診断は人体に対して非侵襲であり安全性が高い。超音波画像の取得には、検者である医師や検査技師の熟練した高度な技術が必要である。検者の技量に依存しない客観的な画像を自動取得できれば、経験の浅い検者でも良質な画像で診断できる。超音波診断が受診しやすくなり、検診の回数を重ねても人体に対して安全な乳がん診断を普及させ、がんを早期発見できる可能性を飛躍的に向上させる。乳腺腫瘍切除手術に際し、腫瘍の位置や大きさを把握し切除領域を決定するために、手術直前や手術中に腫瘍像の3次元モデルを表示する手術ナビゲーションが求められている。超音波診断装置は小型であり、非侵襲のため安全性が高いという特徴があり、診断の場面で幅広く使用されている。しかしながら、医用超音波画像は音波の干渉によるスペックルという特有のノイズが存在し、空間分解能が低いため関心領域の境界が不鮮明となり、鮮明な3次元像が得られない。超音波画像の2次元断層像から3次元像再構成を行うため、2次元画像の段階でスペックルを低減し、輪郭抽出を行う必要がある。本研究では超音波画像から関心領域の輪郭抽出を行った

  • カラー静止画像圧縮に関する研究開発

    研究期間:

    2000年
    -
    2003年
     

  • 電子透かし技術を応用した画像情報管理システムの開発

    日本学術振興会  科学研究費助成事業

    研究期間:

    2000年
    -
    2001年
     

    宮崎 明雄, 櫻井 幸一, 大濱 靖匡, 鎌田 清一郎

     概要を見る

    本研究では,電子透かし技術による画像情報管理システムの開発を目指して研究を行った.まず,基礎的研究として,電子透かし技術について検討を行い,電子透かし方式の開発と改良を行った.また,これらの研究開発のべースとして,電子透かし方式の性能評価法についても検討を行った.そして以上の研究成果を踏まえて,電子透かし技術による画像情報の保護管理について議論し,電子透かし埋め込み検出システム(アプリケーションソフトウェア)の試作を行った.本研究で得られた研究成果は以下の通り.(1)電子透かし方式の開発と改良:ウェーブレット変換を用いた新しい電子透かし方式を提案し,この方式を画像の部分的改ざんやすり替えなど悪意のある攻撃の有無を検出できるように改良した.また,画像の多重解像度表現を利用して新しい相関利用型の電子透かし方式を提案するとともに,この方式の理論的解析を行い,電子透かしの検出精度を改良した.さらに,コンテンツ配布者の不正に対して安全な電子透かしシステムの開発も行った.(2)電子透かし方式の性能・品質・耐性評価について:周波数領域利用型の電子透かし方式に対して,透かし情報の埋め込み検出システムのモデル化を行った.、そして,このモデルを用いて量子化制御型及び相関利用型の電子透かし方式の性能,品質,耐性について議論し,電子透かし方式の評価方法を提案した.(3)電子透かし技術による画像情報の保護管理について:電子透かし技術を用いて,MPEG動画の再生・コピー制御方式,画像コンテンツの検索・識別方式,及び画像の認証・保護方式を提案した.また,Windows版・静止画用電子透かし埋め込み検出ソフトウェアの試作を行った

  • 空間充填走査を用いたディジタル画像システムの構築

    日本学術振興会  科学研究費助成事業

    研究期間:

    1999年
    -
    2000年
     

    鎌田 清一郎, 片山 喜規, 迫江 博昭

     概要を見る

    次世代のMPEG7などに代表される画像圧縮技術では,単なる圧縮のみの手法ではなく,画像の検索、内容抽出なども効率よく行うことができる技術が求められている.本研究では,空間充填曲線を利用してこのような画像検索に向いた新たな画像圧縮技術の開発を目的とする.昨年度と本年度に得た研究成果は次の通りである.(1)まず,画像上の複数の特徴的領域(例えば,平坦な領域)を充填しながら連続的に走査するような新たな空間充填走査方式を実現した.本手法は,画像からMinimum Spanning Tree(最小全域木)に基づく任意形状分割木を作成し,この任意形状分割木を用いて複数の領域をそれぞれ充填しながら連続的に走査するアルゴリズムである.また、この走査情報を最小全域木の枝刈りによって効率よく削減する方法を開発した。(2)次に,トレリス符号量子化手法を用いることにより符号量の削減を実現し,圧縮効率の向上を図った.トレリス符号量子化とは,トレリス符号変調の符号化構造を圧縮に応用したものである.実際の画像を使った圧縮実験では,SN比において,約10分の1の圧縮率で0.5〜1.0dBの向上が見られた.(3)さらに,濃淡画像、カラー画像を対象として,色空間を含めた空間充填走査による画像圧縮手法の圧縮効率と処理速度との関係を,従来手法と比較して明らかにした.JPEGとの比較実験を行い,画像によっては,本手法が優れたり,あるいはやや劣ったりするが,視覚的にはほぼ近い画質であった.しかし,本手法の圧縮処理にかかる時間は,JPEGよりも約20倍かかることがわかった.計算量削減は今後の課題である

  • 2次元パターンのワープ法に関する研究

    日本学術振興会  科学研究費助成事業

    研究期間:

    1998年
    -
    2000年
     

    迫江 博昭, 内田 誠一, 片山 喜規, 鎌田 清一郎

     概要を見る

    2画像のピクセル間の最適マッピングを動的計画法で行う2次元ワープ法に関して検討して,以下の成果を得た.1.2次元的な単調連続性拘束によりトポロジーを保存する基本的なアルゴリズムと,それを改良した高速アルゴリズムを開発した.2.高速化のため区分線形近似アルゴリズムを開発した.3.これらをオフライン文字認識に適用し,基本的な有効性を確認するとともに,問題点を明らかにし,歪み量の事後評価による解決指針を得た.4.不特定話者音声認識への適用を検討し,フレーム間の連続性を保存する周波数ワープを検討して有効性を確認した.5.関連して,雑音画像中の軌跡群検出,オンライン文字認識,画像パターンの捜査方式に関して検討した.以上の成果を,学会誌論文11件,紀要論文4件,国際学会論文5件に発表した

  • 仮想空間構築における実動画像情報を用いたモデル自動生成の高品位実時間処理の研究

    日本学術振興会  科学研究費助成事業

    研究期間:

    1997年
    -
    1999年
     

    谷口 倫一郎, 日下部 茂, 菅沼 明, 鎌田 清一郎, 鶴田 直之, 有田 大作, 吉田 紀彦, 行場 次郎

     概要を見る

    本研究の主な研究成果は以下の通りである.1.仮想空間と実空間の実時間融合のプロトタイプシステムの開発本研究では「仮想空間内に提示するために必要な情報=事前情報+獲得情報」という枠の中で,1.必要以上に情報獲得が難しくならない,2.手軽に事前情報を与えることができる,3.非剛体・多関節体(動物)を取り扱う,という視点らかモデル生成の基本的な枠組みを検討した.この思想に基づき,ここでは,Analysis by synthesis法(Analysis by synthesis:パラメータに基づいてモデルを画像に写像し,その結果が入力画像に最も近くなるようにパラメータを修正する方式)に基づき,以下のような基本アルゴリズムを開発し,評価した.なお,対象物を表現する幾何モデルとしては、「曲げ」や「先細り」を表現することが可能な変形可能超2次曲面(Deformable Super Quadrics;DSQ)を用いた。・初期3次元形状モデル獲得のための対象物モデル作成システム・Analysis by synthesis法による多関節物体の追跡システム・獲得した形状パラメータの視覚化とテクスチャマッピング2.処理の高速化、高精度化に関する検討1.で開発したアルゴリズムを更に改善するために、以下の点について研究を行った。・システムを線形化し、時・モデル空間勾配法により処理速度の高速化する手法を開発した。・多視点画像の中から、解析に最適な視点を選択する手法を開発した

  • 英語・日本語会話文のデ-タベ-ス化と意味情報による検索利用システムの研究

    日本学術振興会  科学研究費助成事業

    研究期間:

    1990年
    -
    1991年
     

    河口 英二, 鎌田 清一郎, 石川 聖二, 石原 好弘

     概要を見る

    1.英・日会話文の収集と意味構造記述デ-タの作成本研究全体を通じて、NHKの「ラジオ英会話テキスト」を原資料として利用し、総数約20,000文の英語/日本語会語文を収集してMSーDOSファイルとした。また意味構造デ-タとしては、約8,000文について作成した。2.会話文と意味構造デ-タ(SD式デ-タ)管理システムの開発SD式処理プログラムパッケ-ジ(SDENV)を開発し、これを基にして「会話文デ-タ管理システム」及び「SD式管理システム)を作成した。その機能としては、A.概念ラベルの登録、B.知識デ-タの動的登録、C.話題(会語場面)管理などを含む。3.会話文デ-タベ-ス利用システムの開発本研究で収集・構築した会話文デ-タベ-スを利用する「英・日会話文の意味的検索システム」のプロトタイプを作成した。このシステムの処理速度を向上させる技法として、処理における以前の成功/不成功結果をシステム内に効率よく蓄積する方法を開発した。この技法により以前の10倍程度の高速化が実現できている。プログラムはPrologおよびCを用いて書いている。本検索システムにおける検索キ-としては、(1)一つのSD式、(2)陳述文における機能項目、(3)概念ラベル、および会話文相互の意味的な近さの尺度を計算して最適なものを検索出来るようにしている。この処理の中心はSDENVの中の「詳述量」とそれを基にした「意味差の尺度」の計算機能である

  • リモ-トセンシング画像を対象とした画像理解に基づく知的インタ-フェイスの研究

     概要を見る

    本研究では、リモ-トセンシングを対象とした画像理解に基づく知的インタ-フェイスに関する研究を行った。1.LANDSATを対象とした意味体系の整理本課題はLANDSAT画像の意味構造を予め貯蓄しておき、入力画像に対するシステムの認識・解釈の曖昧さをなくすため、全体としての処理効率を向上させることにある。そのための意味構造の記述方法としては、本研究代表者等がすでに開発してきたSD式を採用することとした。本年度は前年度までにPROLOGで試作したSD式処理システムの改良を行い、意味デ-タの検索能力を高めた。2.リモ-トセンシングデ-タに関する画像処理アルゴリズムの開発画像の認識・理解では前処理として領域分割、線状物の抽出の問題、さらには大量画像デ-タの蓄積保存等がある。本年度は“複雑さの尺度"を基にした領域分割、部分的画一化処理による画像デ-タの情報圧縮、ニュ-ラルネットによるLANDSAT画像の領域の分類実験等を行った。ニュ-ラルネットによる領域の分類実験では5×5画素の小領域をその濃度値の分布パタ-ンにより5種の領域に分類する実験を行った。その結果、さらに多種類の領域分割にもこの手法が効率良く動作する見通しが得られた。3.知識ベ-スを利用する画像処理システム「IPSSENS」の改良本研究グル-プで従来から開発してきた画像処理支援システムであるIPSSENSの改良を行った。改良の要点は(a)画像処理アルゴリズムの適用法に関する知識をシステムに入れる。(b)専門の画像処理技術者の知識をシステムに容易に取り入れることができる。(c)画像デ-タと属性デ-タが共にシステムで扱える。(d)画像処理の履歴情報をシステム内に残せるようにしたことである

  • SD式モデルに基づく英日会話文デ-タベ-スからの背景知識と意志や意図情報の抽出

     概要を見る

    1.会話文デ-タの収集と意味構造デ-タの蓄積既に蓄積している会話文に加え,新たな会話文テキストのMSーDOSファイル化と意味構造のデ-タを作成した。SD式として新たに蓄積したデ-タ量は約7000文程度である。2.SD式処理システム(SDENV)の改良従来からPrologを用いてSD式実験プログラムパッケ-ジ(SDENV)の高速化と高機能化を図った。高速化の主な手法は、探索処理において以前の成功結果を効率的にシステム内に保存しておき,後の処理で再利用することである。この結果、例えば二つの概念の最近共通先概念を求める問題では、多くの場合、これまでのものに比べて10ー20倍もの高速処理が可能となった。3.会話文検索システムの試作システムに与えられた知識デ-タを利用してデ-タベ-ス内の会話文を意味情報を用いて検索可能となるシステムを試作した。このシステムでの検索キ-としては、(1)一つのSD式(2)陳述的な会話文における機能項目(3)概念ラベル(4)英語/日本語例文の4種を設定した。(1)はSD式で与えられた発話の意図/意味に最も近い英語/日本語内容の例文をデ-タベ-スから探し出すことである。(2)は検索の際、発話者の意図や内容の主語/述語/目的語等々の意味的な機能項目に関する重みの指定を可能とするものであり、焦点を絞るものである。(3)は概念ラベルのみを指定する場合,(4)は例文と同義のものを探し出す場合である

  • SD式モデルに基づく英日会話文データベースからの背景知識と意志や意図情報の抽出

     概要を見る

    本年度は主として、以下の研究を実施した。(A)SD式作成支援ツールの開発と会話文データの収集会話文データを基にしたSD式作成作業は人手によるが、SD式モデルでは「意味記述の具体的な方法」については定めていない。従って作成者毎に異なった書き方をすることがあるがこのような不統一はなるべく回避した方がよい。そこで本年度は「SD式作成マニュアル」の完成を目指した。また、作成作業の効率化を目指してパーソナルコンピュータ上に「SD式作成支援ツール」構築した。これはSD式としての文型(陳述式と感情式)や、様相、時制、相情報等を統一的に作り易くする他、構文的な誤りを防ぐ効果がある。(B)会話文検索システムの改良前年度試作した検索システムには、日本護を入力キーとする検索モードを備えていなかった。本年度は制限された簡単な構文を持つ疑似的な日本語(単語としてはなるべく概念ラベルをそのまま用いるようにしている)をキーとして検索出来るモードを

  • SD式モデルに基づく英日会話文データベースからの背景知識と意志や意図情報の抽出

     概要を見る

    本年度は主として,以下の研究を実施した。(A)SD式作成マニュアルの改訂前年度に編集したSD式を作成するための手引き「SD式作成マニュアル」を再検討し、会話文の実例に即していくつかの改訂を行った。その主な点は、発話意図の分類を詳細にしたことによる発話意図ラベルの追加、及び、結合子"asas","asif","evif"などの新設である。マニュアルにはこれらの用法を加えるとともにSD式モデルの全体を分かりやすいようにする工夫をした。(B)実験環境プログラムパッケージ「SDENV」の改良従来のSDENVでは、詳述スコアの値の設定に関する原理が曖昧であった。本年はこの点を再検討した。この結果、知識に基づく多段推論(アブダクション)の有効段数を3段程度に設定し、この際の詳述スコアと最も簡単な構文的詳述スコアが同程度であるように設定することとした。このことをSDENVに反映させ、詳述量の計算と最近共通先祖の探索アルゴリズムを簡潔なものに出来た。新たなSDENVは「SDENV-」と改名した。C)英日会話文データベースの拡充従来から続けている会話文の収集とSD式データの作成作業を引き続き実施した。原データはこれまで通りNHKラジオ英語会話テキストであった。D)SD式意味構造記述モデルの他分野への応用の検討SD式は自然言語の文の意味構造を記述することを目標に導入してきたものであるが、一般には知識システムにおける知識データの記述法としても有効である。このことを実証することを目的に、環境データ(大気汚染、海水の汚染等)を解析する際の知識表現形式としても利用可能である。本年度はLANDSATデータの解析への応用法の検討を開始した

  • ヒルベルト曲線による走査アドレス・ジェネレータの開発

     概要を見る

    本研究の目的は自己相似性を有する空間充填曲線の一種であるヒルベルト曲線の走査アドレス発生のハードウェア化である。ヒルベルト曲線が自己相似性という興味深い性質をもっているにも拘わらず、その走査アドレス発生に時間がかかるという問題があったが、本研究によりルックアップテーブルを利用した高速計算法のハードウェア化が実現可能となった。本研究ではまず、空間次元数2、3、7におけるハードウェアを試作し、走査時間を計測した。(1)2次元空間では画像サイズが256×256画素、基本クロックが1MHzの場合、すべての画素の走査アドレス発生に要する時間が約0.039秒であることを確認した.ハードウェア規模は2進カウンタとルックアップテーブルに相当する小記憶容量のROMを使用することにより、ICとROMの個数は約15個程度で済み、極めてコンパクトに回路構成できることを確認した。また、ROMの部分が論理回路で簡単に設計できることも確認した。(2)3次元空間では256^3空間画素の場合を設計し、1走査に対して約9.39秒であることを確認した。また、ハードウェア規模はICとROMの個数でみると、2次元の場合より若干増えて約20個程度であることを確認した。(3)7次元空間では256^7空間画素の場合を設計し、1走査に対して計算時間が約4.03×10^<10>秒となり、次元数に対して指数関数的に計算時間が増大することを確認した。次に、本手法では並列計算が可能であるため、その並列ハードウェアを試作し、並列度に対する計算時間を計測した。実験の結果、前記(1)と同じ条件に対して並列度2の場合には約0.019秒となり、約1/2の計算時間になった。また並列度4の場合は約0.012秒となり、若干の計算速度向上が図られた。本研究成果により、これまでのラスタ走査の画像通信に対して、圧縮効率の良いヒルベルト走査による画像通信の構築が可能となった

  • ヒルベルト走査による画像表示装置の開発

     概要を見る

    今年度に行った研究によって、得られた成果は以下の通りである。(1)ヒルベルト曲線は自己相似性という興味深い性質をもっているにも拘わらず、その走査アドレス発生に時間がかかり、ハードウェア規模もラスター走査などに比べ格段に大きくなるため、走査方法として不適とされてきたが、申請者が提案したアルゴリズムを基にして、ヒルベルト曲線による走査アドレス発生のハードウェア化を検討し(論理設計、レイアウト配置)、実際に小規模のLSI(52ピン用)を試作した。(2)(1)の成果を生かし、液晶ディスプレイを使ってヒルベルト走査による2値の画像表示装置を実際に試作した。(3)ヒルベルト走査によるランレングス符号化を基にして人間の視覚特性を利用した濃淡画像圧縮技術について検討した。これは、ヒルベルト走査を適用した画像データベースを想定した場合、自己相似性を有するため、対話型環境に適した画像検索ができるという特徴がある。また、ラスター走査に比べて近傍情報をより保存するため、圧縮効率が良くなっており、濃淡静止画像の圧縮率が標準画像GIRLに対して0.49bit/pixelの画質を標準化手法のJPEGと比較した場合、ほぼ同等の画質であることを確認した。本手法は符号化処理においては JPEGとほぼ同じ処理時間であるが、複号処理にほとんど時間がかからず、JPEGの約1/100で済むという結果を得た。本手法は従来手法に比べて比較的簡単な処理で済むため、小規模のハードウェア構成で実現できるという見通しを得た。(4)本研究により開発した濃淡画像圧縮手法をカラー画像にも適用できるように本手法の拡張を行った

  • ヒルベルト走査を利用したディジタル画像システムの構築

     概要を見る

    カラー動画像を対象として,3次元近傍での高い相関性を有するヒルベルト走査を利用した画像圧縮手法を開発した.これは,ヒルベルト走査に沿って得られた1次元データに対して,線形近似符号化を行うものである.今年度の研究成果としては以下の通りである.(1) 本手法の圧縮効率を検証するためにMPEGとの比較実験を行なった.実験に用いた画像は,カラーディジタル標準動画像MissAmerica(360×288画素,24(bpp),0〜35フレーム)である.本手法の有効性を確認するため,SN比とビットレート,処理速度の関係を明らかにした.また,本手法は圧縮処理において再帰的処理を省くことにより,さらに高速化を図ることができた.この場合,復元画像の画質はSN比に関して0.7(dB)程度低下するが,処理時間はMPEGの約1/10となった.(2) (1)により開発したヒルベルト走査による画像圧縮手法を利用して,実際にディジタル画像システムを構築した.これは,2台のパーソナルコンピュータを使い,符号化におけるCCDユニット・入力回路による画像取り込み,復号化における液晶ユニット・出力回路による画像表示,また,走査アドレス発生回路のハードウェア全体の通信制御を行うものである.走査アドレス発生回路は,LSI化を検討し,サイズ512x512までの画像を走査できるLSIチップを試作した

▼全件表示

Misc

  • 離散コサイン変換に基づく定数時間ガウシアンフィルタの包括的性能解析 (画像工学)

    杉本 憲治郎, 京地 清介, 鎌田 清一郎

    電子情報通信学会技術研究報告 = IEICE technical report : 信学技報   117 ( 48 ) 19 - 24  2017年05月

    CiNii

  • 色クラスタに着目したカラー画像向けの効率的なバイラテラルフィルタ (ライフインテリジェンスとオフィス情報システム)

    杉本 憲治郎, 福嶋 慶繁, 鎌田 清一郎

    電子情報通信学会技術研究報告 = IEICE technical report : 信学技報   116 ( 220 ) 41 - 45  2016年09月

    CiNii

  • 色クラスタに着目したカラー画像向けの効率的なバイラテラルフィルタ (メディア工学)

    杉本 憲治郎, 福嶋 慶繁, 鎌田 清一郎

    映像情報メディア学会技術報告 = ITE technical report   40 ( 31 ) 41 - 45  2016年09月

    CiNii

  • 線型多様体色記述子の明度変化への頑健性に関する一検討 (画像工学)

    杉本 憲治郎, 鎌田 清一郎

    電子情報通信学会技術研究報告 : 信学技報   111 ( 284 ) 31 - 34  2011年11月

     概要を見る

    線型多様体色記述子(LMCD)の明度変化に対する頑健性を向上させる手法を提案する.LMCDは色ベースの薬剤包装識別のために提案された大局色記述子の一つであり,高速・高精度に色分布を照合でき記述サイズも小さい.しかしながら実環境において重要な照明変化への耐性を持たず,またそれに関する評価実験も十分でない.本稿ではLMCDに明度不変となる正規化を施し,実環境での照明変化に対する頑健性の向上を狙う.異なる照明環境で撮影された画像セットを用いた実験において,提案手法は従来手法を有意に上回る識別率を達成した.

    CiNii

  • 閾値に一致する累積寄与率を持つ部分色空間記述子の構築

    杉本憲治郎, 鎌田清一郎

    画像の認識・理解シンポジウム(MIRU2011)論文集   2011   277 - 282  2011年07月

    CiNii

  • 固有色空間の照合による薬剤パッケージ認識

    杉本憲治郎, 井上昻治, 黒木祥光, 鎌田清一郎

    第73回全国大会講演論文集   2011 ( 1 ) 539 - 540  2011年03月

     概要を見る

    本論文では固有色空間の照合による薬剤パッケージ認識について述べる.薬剤パッケージにとって色は重要なグローバル特徴の一つである.提案法は低レベルな色記述子の一つであり,色分布の固有空間を表す.従来の低レベル色記述子である色分布ヒストグラムや色クラスタと比較して,パラメータ設定が不要でありサイズも小さいといった利点がある.性能評価実験では,従来法より高速処理・高認識率であることを確認した.

    CiNii

  • 平均予測値との差分による予測器選択を用いた可逆画像圧縮

    杉本 憲治郎, 黒木 祥光, 鎌田 清一郎

    映像情報メディア学会技術報告   32 ( 54 ) 65 - 69  2008年11月

     概要を見る

    予測符号化において,局所的な指標に基づき予測器を切り替えることで高効率化を狙う手法が数多く提案されている.局所的な指標として隣接画素間の差分(勾配)を用いた手法がMEDやGAPをけじめ多くあるが,勾配と予測器間の関係についての定量的な議論は少ない.本稿では局所的な指標として平均予測値との差分を用い,その指標と予測器間の関係の定量的な議論を試みる.またその議論を基に設計された予測手法およびスキームを提案する.性能比較実験では,提案予測手法はGAPに比べ予測誤差エントロピを0.070[bits/pixel]減少でき,提案スキームはCALICに比べて平均符号長を0.016[bits/pixel]削減できた.

    CiNii

  • 複数の直交基底を用いた最近傍コードワードの高速探索法

    黒木 祥光, 高橋 幸太郎, 上繁 義史, 鎌田 清一郎

    電子情報通信学会技術研究報告. IE, 画像工学   106 ( 397 ) 29 - 34  2006年11月

     概要を見る

    ベクトル量子化では,最近傍コードワードの探索に多くの計算量を必要とするため,高速化のアルゴリズムが必須である.高速探索のアルゴリズムは数多く提案されてきたが,それらは全て理論的に最近傍の候補となり得るか否かを何らかの尺度に基づいて判定し,成り得ないコードワードに対するL_2ノルムの計算を省略するものである.本論文では,先ず,既存の手法であるEENNS(equal-average equal-variance nearest neighbor serarch)法とDHSS(dynamic hyperplane shrinking search)法の関連性について議論する.また,その議論を反映し,WH(Walsh-Hadamard)基底と標準基底の双方を用いた高速化アルゴリズムを提案する.提案法では複数の基底成分を判定に用いるため,条件分岐処理の増加が欠点として考えられる.その対策として確率論的に各基底成分の分散を導出し,高速化に寄与しないと思われる基底を求めた.また,計算機実験との整合性について記した.しかしながら,標準基底の追加による優位性を示すことは出来なかった.

    CiNii

  • A MODEFIED METHOD OF ADAPTIVE SPACE-FILLING CODING

    Ueshige Yoshifumi, Hiratsuka Sei-ichiro, Kamata Sei-ichiro

    電子情報通信学会技術研究報告. IE, 画像工学   103 ( 539 ) 71 - 74  2004年01月

     概要を見る

    This paper describes a new method of adaptive space filling coding. One-dimensional (1-D) pixel data along with a space filling scanning can be generated adaptively from an original image by the conventional method. Because the scanning path is irregular, encoding cost for the scanning path information becomes expensive. This paper presents an algorithm for encoding scanning path information. Through the construction of a minimum spanning tree (MST) to represent the scanning path, absolute sum of difference between adjacent small blocks is evaluated. Context of the image selectively changes in accordance with this evaluation in order to generate Hilbert tree as a regular tree in flat region. Reduction of this scanning path information can be achieved. It is sufficient for describing the MST to encode the difference between the MST and the regular tree. Our experiment results show that the proposed method is efficient.

    CiNii

▼全件表示

産業財産権

  • 薬剤判別用撮影装置及び薬剤判別用写真撮影方法

    鎌田 清一郎

    特許権

  • 画像フィルタ演算装置及びガウシアン・カーネル演算装置並びにプログラム

    鎌田 清一郎, 杉本 憲治郎

    特許権

  • 暗号処理システム、暗号化装置、復合装置、及びプログラム、並びに暗号処理方法

    鎌田 清一郎

    特許権

  • 相同性検索装置及びプログラム

    鎌田 清一郎, 唐 海江

    特許権

  • 画像識別装置及びプログラム

    5582610

    鎌田 清一郎, 杉本 憲治郎

    特許権

  • 粒状物品種検査装置

    5163985

    鎌田 清一郎

    特許権

  • 認証装置及び撮影装置

    鎌田 清一郎

    特許権

  • 指紋特異点抽出装置及び指紋特異点抽出方法

    鎌田 清一郎, 許 霄

    特許権

  • 画像変換パラメータ演算装置及び画像変換パラメータ演算方法

    鎌田 清一郎, 田 黎

    特許権

  • 予測器及び画像符号化器

    鎌田 清一郎, 唐 海江

    特許権

  • 画像検索方法及び画像検索装置ならびにプログラム

    鎌田 清一郎

    特許権

  • マッチング方法およびマッチング装置ならびにプログラム

    4570995

    鎌田 清一郎

    特許権

  • 硬貨識別装置および硬貨識別方法

    4444089

    鎌田 清一郎, 唐 海江

    特許権

  • 画像検索装置及び画像検索方法

    鎌田 清一郎

    特許権

  • ヒストグラム近似復元装置及びヒストグラム近似復元方法、並びに画像検索装置及び画像検索方法

    4575751

    鎌田 清一郎

    特許権

  • 画像符号化装置及び画像符号化方法

    鎌田 清一郎

    特許権

▼全件表示

 

現在担当している科目

▼全件表示

 

他学部・他研究科等兼任情報

  • 理工学術院   基幹理工学部

学内研究所・附属機関兼任歴

  • 2022年
    -
    2024年

    理工学術院総合研究所   兼任研究員

特定課題制度(学内資金)

  • カラースペクトル画像解析による眼底年齢推定に関する研究

    2022年  

     概要を見る

    本研究では、白内障、糖尿病、緑内障、網膜症などを対象としたカラースペクトル眼底画像を用いて、ResNetやEfficientNetなどの深層学習による眼底年齢推定の研究を行った。具体的には、これまでに行ってきた糖尿病網膜症における識別法を拡張して、どのような特徴が識別および推定に有効かを調査しながら、眼底年齢推定を行った。北京大学データセットによる実験の結果、正常な眼底画像における識別精度の平均二乗平方根誤差は2.5歳となり、先行研究より優れていることが示された。また網膜出血、微小動脈瘤などの血管や視神経乳頭の症状が年齢推定に影響を及ぼしていることが示唆された。本研究成果の一部は、IEEE&nbsp; ICIP2022において研究発表した。

  • 深層学習によるER-IHC画像の癌細胞検出に関する研究

    2021年  

     概要を見る

    癌細胞画像などによるスクリーニングには、増大する検査量に対して過度の負担が細胞検査士にかかっている。本研究は、当該検査士の負担軽減のため、Estrogen Receptor-Immuno HistoChemistry (ER-IHC)画像を対象とし、深層学習によるER-IHC画像から癌細胞の高精度検出する方法論の確立を目的とした。この目的を達成するために、UnetモデルとRefinement moduleを融合した2段階Refinementネットワークを開発し、新たな指標による癌細胞検出アルゴリズムを確立した。マレーシアのマルチメディア大学との共同研究で使用しているER―IHC画像データセットを利用して評価実験を行い、その有効性を確認し、小細胞や細胞核境界がぼけた場合などに特に有効であることがわかった。&nbsp;

  • 多細胞画像の核領域分割および検出に関する研究

    2020年  

     概要を見る

    病理診断においてHematoxylinやEosin染色画像は、H&E画像とも呼ばれ、最も広く利用されている。本研究では、細胞核の状態により癌レベルの自動判定支援を行うことを目的として、細胞核の大きさ、密度、色などの分布の尤度に基づき、Detect-U-Net、境界-Segment-U-Net、Regression-U-Netという3種類のモデルを有機的に結合させた、H&E画像からの細胞核中心とその領域の予測方法を開発した。胸、肝臓、腎臓などの9種類の臓器の細胞画像データセットを使って従来手法と比較した結果、細胞核検出および領域分割の精度が約10%~15%向上し、多臓器間の高精度細胞核検出と領域分割に有効であることがわかった。また、本研究成果を国際会議IVPR2020において発表し、Awardを受賞することができた。

  • 糖尿病網膜症のための眼底画像識別法に関する研究

    2019年  

     概要を見る

    糖尿病網膜症は失明する原因の多くの割合を占め、早期発見と適切な治療で視力障害を防ぐことが可能である。本研究は、診断支援のための疾患重症度評価に基づいた高精度な眼底画像識別法の確立を目的とする。この目的を達成するため、新たなグラフ表現を導入したグラフニューラルネットワーク(GNN)の適用を検討した。これは、病変領域を抽出後、その依存関係をグラフで表すことにより病変領域依存グラフを構築し、病変領域の依存関係にはGNNを用い、特徴抽出CNNと融合した眼底画像識別法を提案した。眼底画像データセットEyePACSなどを用いて従来手法との比較評価実験を行った結果、本提案手法が識別精度面で数パーセント向上することを確認した。

  • 多次元医用画像の高効率解析アルゴリズムに関する研究

    2017年  

     概要を見る

    脳科学分野などでは画像データの高密度化・高次元化が顕著であり、ハードウェアの進歩だけでは実時間処理が困難な現状がある。本研究は、多次元画像を対象とし、高速フィルタ処理、高速特徴抽出法などのアルゴリズムの開発を行った。まず画像処理における基本的処理として用いられるバイラテラルフィルタ設計に関して、特異値分解を用いた効率的な手法を提案した。従来法と同等の近似精度を維持し、およそ5割少ない計算量で実現した。次に、病理組織の染色画像を対象として、Secant Normal Votingに基づく細胞のセグメンテーション手法を考案した。本手法は、従来問題となっている重複領域などに対しても良好な細胞核検出が実現できることがわかった。

  • 悪天候下の深層学習による道路環境の画像認識技術に関する研究

    2017年  

     概要を見る

    高齢者用の運転支援などを実現するため、エコ電気自動車の周囲環境認識技術について研究している。本研究では、まず車体周囲の3次元環境情報を取得する上で、ステレオビジョンの計測精度を高めることを検討した。これまで奥行き情報を取得する上で物体間の境界がぼけてしまうという問題があったが、これを上下左右に延びたクロス形状のフィルタを設計し、これを2段階で適用する新たな方法論を確立した。次に、悪天候下や夜間走行を行う上で近赤外カメラも利用して自動車周辺の状況把握に深層学習を利用することを検討した。その結果、エッジがクリアに再現され、安定したカラー領域が生成される新たな近赤外カラー化モデルSNetを開発した。

  • 多次元高精細バイオ画像データの高速画像処理アルゴリズムに関する研究

    2016年  

     概要を見る

    多次元高精細バイオ画像データにおけるアライメントを含む高速3次元再構成法について、アルゴリズム面での改善を含めて実時間の処理が可能となる手法を研究開発した。まず核磁気共鳴画像法(Magnetic Resonance Imaging)による3次元ボリュームデータのアライメントに対して、高速かつ高精度な画像処理アルゴリズムを確立した。従来手法と比較して数%~10%の高速化を図ることができた。また、肝臓などの細胞画像の3次元アライメントによる画像再構成アルゴリズムを確立し、従来手法に比べて高精度化および高速化を図ることができた。

  • 読み聞かせロボットの高精度画像認識に関する研究

    2013年  

     概要を見る

    二宮金治郎をイメージしてデザインを考え、薪のかわりにパソコンを背負うというコンセプトにより、感情を込めて本の読み聞かせを行うロボット「二宮くん」を開発している。「二宮くん」は、目にカメラを搭載しているため、一般照明の影響、本のフォントの多様性により文字認識がうまくいかず、認識率があまり良くないという問題があった。本研究では、「二宮くん」の高精度認識を実現することを目的としてこれらの問題を解決するため、次のような研究成果を得た。(1)照明等による影響による画像の変動を吸収するために、カラーコントラスト補正法を開発した。これはRGBカラー空間の輝度情報に着目し、大域処理と局所処理の2段階の階調マッピングを実現する階層的ヒストグラム表現を用いるものである。NASA Langley Research Centerのデータセットを使った評価実験の結果、提案手法は代表的な従来手法よりPSNRによる評価において約10%、エントロピー評価においても同等の画質改善ができることを確認した。(2)本などの画像から文字部分を高精度に抽出し認識する必要がある。本研究では、Gaborフィルタを拡張し、新たに局所Gaborテクスチャ特徴を提案した。これは画像中における局所領域のコーナー点の周辺の色とテクスチャを用いた特徴量である。色情報はHSV色空間から得られる画像中の色彩と色相の情報から抽出し、テクスチャ情報は明度情報からガボールフィルターを用いて抽出する。また、テクスチャ情報は離散フーリエ変換を適用し、画像の回転に対して不変の値にすることで、画像中のテキストにおいて頑健な特徴量を抽出した。文字認識・文書理解国際会議にて一般公開されている画像データセットを用いてテキスト領域の検出性能を評価した場合、提案手法の性能は他手法の性能よりも再現率において約10%、F値において数パーセント高いことを示した。これらの研究成果により、独自に収集したデータセットを使った認識実験の結果、従来の文字認識法に比べて数パーセントの認識精度の向上を図ることができた。

  • コミュニケーション・ロボット「二宮くん」の研究開発

    2010年  

     概要を見る

    2009年6月図書館、介護施設等の本の読み聞かせを代行あるいは支援してくれる、本の読み聞かせロボット(ブック・リーダ・ロボット)「二宮くん」(にのみや・くん)を試作した。「二宮くん」は、これまでにない、前に置かれた本を読みあげてくれる、親しみのある人型ロボットである。これまで「二宮くん」の展示会出展などを通して、用途が本の読み聞かせのみではなく、コミュニケーション・ロボットとしての役割を果たすともっと面白いロボットになるのではないか、という意見を多数頂いた。本研究では、このようなニーズに応えるためにコミュニケーション・ロボットとしての機能を実現することを目的とする。2010年6月にも、研究開発を行ってきた本の読み聞かせロボット「二宮くん」を展示会に出展した際に、歌を歌わせること、対話ができること、という要望が強かったが、今年度はまず、本を読むだけでなく、楽譜を読ませる機能を追加して、歌を歌わせることを試みた。画像処理により、五線譜部と歌部の区別を行い、ドレミの音符の認識を行った。その結果、音符は90%以上の精度で読みとることができた。歌詞の認識も行う必要があり、音符と連動できるようにした。次に、家庭という空間を想定した場合に、本などをベースにしたコミュニケーションを行うことを念頭に、顔認識機能の追加を行った。その結果、複数走査によるパターンコード表現に基づいた顔認識アルゴリズムを確立した。公開されているORLデータ、FERETデータなどの様々な画像データに対して顔認識のアルゴリズムを評価した結果、従来手法LBP、LDP等に対して数%の認識率向上が得られた。また、音声認識方式の開発、対話テキスト解析方式の導入を検討しているが、今後コミュニケーション・ロボットのプロトタイプシステムを構築し、人間と比較した評価を行うことによって改善を図る。

  • 空間充填曲線による画像圧縮・検索に関する研究

    2008年  

     概要を見る

    本研究では、これまでに構築してきた空間重点曲線を用いた画像圧縮法に対して、計算効率のよい画像検索を行うことを目的として検索手法を開発した。これまでの画像圧縮法によって圧縮されたデータ(以下、圧縮データ)から探索対象物(例えば人物、車など)を何らかの特徴で表現できる高速画像検索の実現を目指して、空間充填曲線を用いた画像検索方式を確立している。この特徴は、圧縮データから空間情報が抽出できるので、様々な記述子を用いて表現することができる。本研究ではフーリエ記述子とPCA(Principal Component Analysis)を用いた新たな方式を提案した。画像MPEG7の形状データベースを用いて、MPEG7に採用されているCurvature Scale Space記述子、Wavelet記述子等の従来手法との類似検索などの比較を行い、その検索効率をRecall-Precision指標等により明らかにした。次に、画像上の各局所領域を予測符号化に適した画像情報のコンテキストモデル化を考察した。コントラストがありエッジが保存できれば、その画質はくっきりした明瞭性に優れたものとなる。すなわち、高コントラスト、エッジ保存型であり、平坦な部分は滑らかに変化するようなモデル化を行うことが必要となる。各局所領域に対して、予め与えられた単純予測器を適応的に選択することによって圧縮効率を上げるものである。これにより、国際標準化方式と比較した評価によって、本手法の有効性を明らかにした。画像検索手法は、蓄積した膨大な圧縮データに対して検索を行うため、特に高速化が要求される。このため、本研究では、FPGAを利用して、高速アルゴリズムを実現し、ソフトウェア性能よりどの程度高速化されるかを明らかにした。

  • 空間充填曲線による画像情報のモデル化と画像圧縮・検索への応用

    2007年  

     概要を見る

    生活・安全などにおいて、画像の圧縮伝送・蓄積に関する研究開発が社会的なニーズとともに益々重要となっている。例えば、遠隔監視では、セキュリティ対象画像を24時間圧縮伝送し、記憶装置に大量の圧縮データを蓄積させている。しかし、これに関わる研究開発者は国際標準化方式である MPEG (例えばMPEG4 AVCなど)などを利用しているが、(a)離散コサイン変換(DCT)、動き補償などを利用した低域周波数通過型の圧縮方式であるために画質がぼける、(b)検索したい人物画像をMPEG圧縮による蓄積データの中から高速に検索できない、(c)圧縮データフォーマットが公開されているための情報漏洩、などの問題が指摘されている。本研究では、これらの問題を解決すべく、MPEGとは異なる、空間充填曲線を用いた画像圧縮方式を検討し、高速検索の実現も含めて、画像圧縮技術の高画質化、高解像度化、高速検索が可能なものを実現することを目的とする。本研究によって、具体的に次のことが明らかになった。(1)画像上の各部分領域を任意形状とし、空間充填曲線を用いて圧縮に適した画像情報のモデル化を行った。これは、空間充填曲線を複数組み合わせることによって、輪郭情報保存性のみならず一般人の主観的画質評価に優れた新たな圧縮手法を開発した。本手法は、独自圧縮方式であるため、圧縮データフォーマットを公開しなければ(c)の情報漏洩の問題は一部解決できるが、セキュリティ面でより頑健なものにするために走査法に着目した暗号化についても検討した。(2)蓄積した膨大な圧縮データからの特徴抽出および画像照合を行い、人物などの類似画像を高速に検索する画像検索手法を開発した。色情報等に基づくヒストグラムによる検索によって、数十倍の高速化が可能となった。なお、MPEGでは圧縮データが周波数領域での表現になっており、DCTなどの逆変換を行う必要があり、高速検索が困難である。

  • 複数走査を用いた画像圧縮および画像検索技術の開発

    2005年  

     概要を見る

    遠隔監視,遠隔医療などにおいて,画像の圧縮伝送・蓄積に関する研究開発が社会的なニーズとともに益々重要となっている.例えば,遠隔監視では,セキュリティ対象画像を24時間圧縮伝送し,記憶装置に大量の圧縮データを蓄積させている.しかし,現状では,国際標準化方式である Motion JPEG,あるいは MPEG などに頼っており,(a)DCT,ウェーブレット変換等の画像変換を利用した低域周波数通過型の圧縮方式であるために画質がぼけてしまうという問題,(b)検索したい人物画像を蓄積データの中から高速に検索できない問題,(c)圧縮データフォーマットが公開されているための情報漏洩の問題,など大きな問題が指摘されている.本研究では,このような遠隔監視あるいは遠隔医療等への応用を目的として,複数の画像走査を利用した,画像変換を行わない高効率な画像圧縮技術と,蓄積した圧縮データから高速に類似画像等の検索を行う画像検索技術を開発した.具体的な成果は,次の通りである.(1) 画像上の各部分領域を圧縮に適した走査方法で走査し,エッジを保存するような画質明瞭性に優れた新たな圧縮技術を開発し,前述の国際標準化方式と比較した定量的評価によって,本開発手法の画質明瞭性を明らかにした. 本手法は,独自圧縮方式であるため,圧縮データフォーマットを公開しなければ情報漏洩の問題は解決できる.(2) 蓄積した膨大な圧縮データからカラーヒストグラムに基づいた画像照合を行い,人物等の類似画像を高速に検索する画像検索技術を開発し,すべての画像の画素単位での類似検索について,その検索効率を明らかにした.(3) 画像圧縮技術および画像検索技術を組み合わせたシステム全体をFPGA上に作成し,本技術のハードウェア規模を明らかにした.

  • 幾何イメージ思考のための支援ツールの開発

    2005年  

     概要を見る

    学生の大半が空間次元拡張をなかなかイメージできないのが現状である。画像処理研究の道具として使っている空間充填曲線(space filling curve、日本ではペアノ曲線とよく呼ばれる)は、曲線による次元拡張の例を示す最もよい道具である。これは、簡単なパターンの回転、縮小、平行移動などの写像を使って興味深い幾何模様が描けるという特長がある。このような現状から、約10年前から自らが翻訳した「空間充填曲線とフラクタル」のテキスト(シュプリンガー・フェアラークから1998年に出版)を使用して、学生に情報数学教育を行っている。図を使うことはイメージを膨らます最もよい方法だと言われるが、空間次元の拡張も同様に思考能力を向上させる重要なテーマである。本研究では、空間充填曲線の生成を基礎として幾何イメージの思考能力を高めるための支援ツールの開発を行った。ペアノ、ヒルベルト、シェルピンスキー、ムーアなどの大数学者が発見した空間充填曲線を使って、2次元平面での生成方法をグラフィックスでわかりやすい教材コンテンツを作成した。次に、これらの曲線を3次元に拡張し、3次元空間における様々な空間充填曲線の模型を作り、3次元立体視表示ソフトとコンピュータグラフィックスを使って2次元から3次元への幾何イメージを思考訓練させるコンテンツを作成した。さらに、n次元空間(n≧4)に拡張したときに幾何図形の認識理解を助ける表示方法を検討した。ただし、表示装置は2次元あるいは3次元の表示系であるため、表示方法は、幾何図形を2次元平面あるいは3次元空間への投影、あるいは、2次元あるいは3次元断面の生成を基本としたものである。

  • 複数の空間充填走査を用いた画像圧縮技術の開発

    2004年  

     概要を見る

    本研究では,予めテンプレートとして与えられた複数の空間充填曲線を利用して新たな画像圧縮技術の開発を目的として研究を行った.具体的な成果は次の通りである.(1)画像上の複数の特徴的領域(例えば,平坦な領域)を充填しながら連続的に走査するような新たな空間充填走査方法を検討し,必要なテンプレートについて圧縮効率と計算量の関係を導いた。(2)静止画像および動画像を対象とし,木構造表現に基づいた画像の階層的分割を行い,個々の分割領域に複数の空間充填走査を適用して画像を表現し,画像記述に適した効率的な画像圧縮技術を開発した.これは,予測係数を逐次更新する予測符号化方式ALCM(Activity Level Classification Model)を改良し,コンテキストモデル符号化と組み合わせた予測符号化方式である.複数のカラー標準画像を用いた圧縮実験の結果、提案手法は国際標準化方式JPEG2000およびJPEG-LSを越える圧縮性能を得た.また,提案手法は,符号化・復号化処理において国際標準化方式に比べて比較的簡単な処理で済むため,ハードウェア規模がこれらの国際標準化方式に比べて小さく実現できることを確認した.(3)近年標準化されたJPEG2000や次世代のMPEG7などに代表される画像圧縮技術では,単なる画像圧縮のみの手法ではなく,画像の検索,内容抽出なども効率よく行うことができる技術が求められているが,圧縮データストリーム上での部分画像検索が行うことができない.そこで,本研究では圧縮データストリーム上で部分画像検索を行うことができる圧縮方式を検討した.今後は,画像圧縮技術および画像検索技術を組み合わせたシステム全体を構築し,本技術の検索性能およびハードウェア規模を明らかにする予定である.

  • ヒルベルト曲線を利用した多次元データ管理システム

    2003年  

     概要を見る

     本研究では,画像表現形式のまま移動や拡大等の処理ができ,範囲検索が高速にできるような多次元データ管理手法として、ヒルベルト曲線を用いて線形4分割木構造表現を行う手法を検討した.これは,空間内においてデータの有無を1,0の2値データで表現し,データがある領域をアドレスと領域サイズの組を一つのデータとして表し,そのデータの組をヒルベルト曲線が辿る順に並べて、多次元データを表現する手法である.ヒルベルト曲線はデータ間の近傍保存性が良く,データのクラスタ抽出ができるという特徴を持つ。範囲検索を高速に行えるMD木による範囲検索では,検索する領域を分割し,それぞれの検索領域を木構造の根から順にノードを辿って検索する.これに対し本手法は,1次元データで管理していることから,検索する領域を1次元データの先頭から検索領域が見つかるまで探索し,検索できれば,分割された他の検索領域は1次元配列上で近傍に存在するはずであるため,その近傍を検索するだけでよい.また,本手法は領域をアドレスで管理していることから,検索はアドレスの比較により簡単に行うことができる.本研究では,ヒルベルト曲線を用いたデータ表現から,表現形式を変えずに,移動,拡大・縮小,回転,抜き取り,共有,交わり,投影,制約,結合,差といった10種の処理を行う手法について述べた.2次元データである2値画像を用いて,本手法の有効性を確認した.また、MD木によるデータ表現において範囲検索の検索速度を比較し、本手法はMD木に対して20~30%検索効率がよいことがわかった.さらに,本手法の応用として,実時間描画のためにデータベースから高速な検索が必要である仮想都市空間のデータ管理への本手法の適用可能性について検討した.今後は,高速に範囲検索が行え,データ表現のまま種々の演算処理ができることから,仮想都市空間のデータ管理への適用をさらに検討する予定である.

▼全件表示