Natural disaster is one of the most important research topics worldwide. In this paper, a data driven cyber-physical system is introduced to detect landslides. This system is composed of Wi-Sun acceleration sensors, which can detect the acceleration of the nearby environment in 3D domain, and the sensors are linked with the router (act as ’sink’ node) via Wi-Sun transmission (i.e. IEEE802.15.4g). The details of the detection system are explained and the landslide detection mechanism with low computational complexity is proposed. A traffic reduction method is proposed thereafter to help reduce the data needed for transmission by exploring the intra-correlations of the sensor data. This method can save the energy consumption without degrading the detection performance. Field test is conducted and the results show that the landslide can be detected and amount of data to be transmitted can be reduced, which verifies the system’s effectiveness.

近年,出力機器における解像度が増加し,より高精細な画像の出力が求められるようになってきている.この高精細な画像を取得する手法として,現在,長年研究されてきた超解像処理技術を符号化技術に適応する手法が研究され始めている.超解像処理技術とは入力された低解像度画像中に欠損した高周波成分を付加することによって高解像度画像を取得する技術であり,中でも学習型超解像は高画質な画像を取得できる手法として知られている.我々はこの学習型超解像であるAplus,ScSR,SRCNNについて方向性を考慮した超解像処理を行い,これらをH.265符号化画像に適応する手法を考案し,符号化画像の更なる画質向上について検討する.

Disaster detection is an important research topic and draws great attentions from both industry and academia. In this paper, we study the hybrid landslide detection system, which utilizes the video surveillance camera and multiple kinds of sensors and can detect the landslide automatically. Edge processing is adopted in this hybrid system to fuse the sensor data based on the Dempster-Shafer (D-S) theory, i.e. utilizing multiple sensors' information to calculate the possibility of the landslide. Edge processing can make faster decisions than the control center, the results of the edge processing are then used to schedule the sensor's transmission frequency and video transmission under the network constraints in this system. The simulation results show that the proposed scheme outperforms the competing schemes in typical network scenarios.

Disaster detection is an important research topic and draws great attentions from both industry and academia. In this paper, we study the hybrid landslide detection system, which utilizes the video surveillance camera and multiple kinds of sensors and can detect the landslide automatically. Edge processing is adopted in this hybrid system to fuse the sensor data based on the Dempster-Shafer (D-S) theory, i.e. utilizing multiple sensors' information to calculate the possibility of the landslide. Edge processing can make faster decisions than the control center, the results of the edge processing are then used to schedule the sensor's transmission frequency and video transmission under the network constraints in this system. The simulation results show that the proposed scheme outperforms the competing schemes in typical network scenarios.

Wireless sensor networks (WSNs) are widely deployed to help detect disaster such as landslides, etc. in countries like Japan due to their efficiency and low cost. But there are various problems in these systems, such as the different kind of noise (such as random visit from animals, which leads to unnecessary false alarms). Orthogonally, context aware activity recognition draws increasing attention recently, where different environmental or martial sensors are adopted for activity recognition in various application scenarios. This paper studies a disaster detection system utilizing the Wi-Sun acceleration sensors together with the Wi-Sun signal fluctuation and considers the detection of human beings in the scene of interest. Hence, the false alarms introduced by the animals' walk in, etc. could be detected using similar method. During the detection process, not only the wireless signal but also the acceleration value collected by the sensors are adopted for a better detection result. Then feature values are calculated for event detection based on the data samples. K nearest neighbors (KNN) is used to classify the events in the scene of interest: empty and non-empty. The detection results are promising as shown in the result section and the proposed method is applicable to a real landslide to avoid animal- induced false alarms.

WSNs are good options to help monitor the scene of interest and notify the unusual happening to control center. But sensors' high sampling rates lead to tremendous network traffic over the bandwidth-limited and energy-critical WSNs, hence how to reduce the network traffic while maintaining the unusual events monitoring function becomes important. In this paper, we investigate the intra-correlations of the data generated by each sensor at different time instances. And we propose a traffic deduction algorithm exploring the sensor data's intra-correlations which could reduce the data volume significantly and guarantee the parameters needed for unusual detection are delivered.

WSNs are good options to help monitor the scene of interest and notify the unusual happening to control center. But sensors' high sampling rates lead to tremendous network traffic over the bandwidth-limited and energy-critical WSNs, hence how to reduce the network traffic while maintaining the unusual events monitoring function becomes important. In this paper, we investigate the intra-correlations of the data generated by each sensor at different time instances. And we propose a traffic deduction algorithm exploring the sensor data's intra-correlations which could reduce the data volume significantly and guarantee the parameters needed for unusual detection are delivered.

In recent years, Japanese comics have been very popular in the world. In this paper, we applied the HOG and AdaBoost algorithm to detect character's face in comic images. We have recognized that learning process requires at least 80 times repeatedly and the detection success rate converges in about 65%.

In recent years, studies of extracting meta-data from comic images has been focused on for query and search application of e-comics. In this paper, we propose to apply Deformable Part Model, which is a detection method using parts placement of an object, to detect character' faces in comic images.

Content cache effectiveness in video distribution is mathematically analyzed. Targeted video distribution models are two types. One is simulcast, the other is scalable approach. Simple two layer model is used for the analysis. Caching effect is studied focusing on the amount of required transmission data based on some assumptions with a certain cache hit probability. The result shows that the scalable approach cannot improve the caching effect.Content cache effectiveness in video distribution is mathematically analyzed. Targeted video distribution models are two types. One is simulcast, the other is scalable approach. Simple two layer model is used for the analysis. Caching effect is studied focusing on the amount of required transmission data based on some assumptions with a certain cache hit probability. The result shows that the scalable approach cannot improve the caching effect.

マンガは 2 値線画からなる複雑な模様によって描かれている．従来，画像のエッジをとらえる画像特徴量として，HOG が使用されており，マンガにも一定の解析結果をもたらすことがわかってきている．その一方で，2 値の線画を扱ううえで，HOG 特徴量では形状をとらえるために冗長な次元がある．また，2 値の斜線は縦横 2 方向の合成によって構成されることから，方向の取得にも難点がある．本稿では我々の提案する 2 値線画像用に作成した特徴量 Histograms of Binary Oriented Pattern(HBOP) を用いてマンガ画像の解析を行う．マンガ画像解析において従来の HOG 特徴量と HBOP を用いた場合の比較結果を示す．Comic image was drawn into complex pattern using binary line. In conventional research, HOG it can treat image edge was utilized for to analyze Comic image as image feature. On the other hand HOG feature has redundant dimension for to treat binary line image. Moreover, diagonal line on binary image was consisted by combination of vertical and horizontal edge. Thus it is hard to get line angle. Histograms of Binary Oriented Pattern(HBOP) was proposed for to analyze binary line image by us. In this research, We analyze comic image using HBOP. Then, comparison results between HOG with HBOP on Comic image analysis was derived by examination.

HTML5 では，ビットマップ形式のデータを扱う Canvas，ベクター形式のデータを扱う Inline SVG がサポートされ，JavaScript ベースで HTML 内に図形や画像を直接記述できる．しかし，Flash，Java などのプラグインを使用した場合と比べ，描画パフォーマンスが劣るという問題点がある．本研究では，Canvas および SVG における従来の描画パフォーマンス高速化手法の比較を行った．さらに，描画する要素に対して最適な手法を自動選択し，適切な手法で描画するアルゴリズムを提案する．In HTML5, Canvas dealing with data of the bitmap and Inline SVG dealing with Data for vector are supported. Thus we can directly describe image or graphic in HTML, by the JavaScript. However, drawing performance is not good compared with the plugins such as Flash or Java. In this paper, we sinvestigate the performance of traditional approaches to increase drawing speed. Finally, we propose an algorithm to select the best approach automatically.

近年，マンガは日本国内のみならず海外においても多数の読者を持つコンテンツとなっている．また，電子書籍閲覧可能な端末の進化と電子書籍市場の発展により，電子化されたマンガの需要も高まっている．マンガは絵，文字，擬音などの多数の要素がすべて一枚の誌面上に描かれるため，大変複雑な構造を持つ．これら登場人物や多数の要素はメタデータとして取り出すことで，ディジタル化されたマンガの利便性を向上させることができる．マンガは基本的に白黒で描かれるため，画像の特性は自然画像と異なる．このため，マンガの登場人物解析を行うためには，マンガの登場人物の特徴をうまく捉える手法を用いる必要がある．本稿では，HOG (Histograms of Oriented Gradient) とSVM (Support Vector Machine) による登場人物の瞳，顔の検出処理と，登場人物の識別処理を組み合わせることで，マンガのページ上から，特定登場人物の検出を行う手法を提案する．結果として，提案手法による特定登場人物の検出で最大0.93 のPrecision を得た．

High computational complexity of a fractional pixel motion estimation (FME) module can in no way be negligible although the module improves visual quality after the integer pixel motion estimation (IME) process. Most of the conventional FME methods must include the interpolation procedure to form fractional pixel search points from the information of the integer pixel matching error costs. The interpolation, however, requires a considerable memory usage and certain amount of processing time. In this paper, therefore, we propose interpolation-free fractional pixel motion estimation techniques by using the data trend approximation and the error cost scaling. The results of our simulation show that the proposed methods produce similar or better performances compared with the existing FME methods, whereas they do not need any additional search points.

In general, a motion estimator in today's video coding includes fractional pixel motion estimation (FME) as well as integer pixel motion estimation (IME). FME can get better quality performance at the cost of higher computational complexity than IME alone. In this paper, a modified parabolic prediction based FME for H.264 video coding is proposed. The proposed method uses specific correction coefficients to improve the PSNR performance of the existing prediction based algorithm. In the simulation results, compared with the original algorithm, our technique shows more accurate predictive power, meanwhile, it does not require the interpolation process.

High computational complexity of a fractional pixel motion estimation (FME) module can in no way be negligible although the module improves visual quality after the integer pixel motion estimation (IME) process. Most of the conventional FME methods must include the interpolation procedure to form fractional pixel search points from the information of the integer pixel matching error costs. The interpolation, however, requires a considerable memory usage and certain amount of processing time. In this paper, therefore, we propose interpolation-free fractional pixel motion estimation techniques by using the data trend approximation and the error cost scaling. The results of our simulation show that the proposed methods produce similar or better performances compared with the existing FME methods, whereas they do not need any additional search points.

Scale-invariant features are widely used for image retrieval and shape classification. The curvature of a planar curve is a fundamental feature and it is geometrically invariant with respect it the coordinate system. The curvature-based feature varies in position when multiscale analysis is performed. Therefore, it is important to recognize the scale in order to detect the feature point. Numerous shape descriptors based on contour shapes have been developed in the field of pattern recognition and computer vision. A curvature scale-space (CSS) representation cannot be applied to a contour fragment and requires the tracking of feature points. In a gradient-based curvature computation, although the gradient computation considers the scale, the curvature is normalized with respect to not the scale but the contour length. The scale-invariant feature transform algorithm that detects feature points from an image solves similar problems by using the difference of Gaussian (DoG). It is difficult to apply the SIFT algorithm to a planar curve for feature extraction. In this paper, an automatic scale detection method for a contour fragment is proposed. The proposed method detects the appropriate scales and their positions on the basis of the difference of curvature (DoC) without the tracking of feature points. To calculate the differences, scale-normalized curvature is introduced. An advantage of the DoC algorithm is that the appropriate scale can be obtained from a contour fragment as a local feature. It then extends the application area. The validity of the proposed method is confirmed by experiments. The proposed method provides the most stable and robust scales of feature points among conventional methods such as curvature scale-space and gradient-based curvature.

The high computation load of fractional-pixel motion estimation (FME) cannot be ignored although it improves visibly the PSNR performance after the integer-pixel motion estimation (IME) process. Most of conventional FME methods must include the interpolation procedure to form the fractional-pixel points from the integer-pixel information. However, the interpolating requires a considerable memory usage and any amount of processing time. In this paper, therefore, we propose interpolation-free fractional-pixel position prediction techniques based on Bezier curve and post the results of the simulation which show a remarkable performance improvement compared to the existing FME methods in terms of PSNR.The high computation load of fractional-pixel motion estimation (FME) cannot be ignored although it improves visibly the PSNR performance after the integer-pixel motion estimation (IME) process. Most of conventional FME methods must include the interpolation procedure to form the fractional-pixel points from the integer-pixel information. However, the interpolating requires a considerable memory usage and any amount of processing time. In this paper, therefore, we propose interpolation-free fractional-pixel position prediction techniques based on Bezier curve and post the results of the simulation which show a remarkable performance improvement compared to the existing FME methods in terms of PSNR.

全変動は輝度変化が滑らかであるという仮定を導入することなく画像を扱える正則化基準である.全変動最小化によって入力画像から骨格成分を取得する手法として,(1)骨格成分とは双対の関係にある振動成分に着目し,半陰的再急降下法により解く手法や,(2)ノイズエネルギーが既知の値以下であるという制約条件付き最小二乗問題と全変動を画像の分類から決定される値にする凸最適化問題の二つを交互に解く手法が提案されている.しかし,これらの手法は計算コストが高いという問題がある.本論文では,全変動最小化を高速に計算する手法を提案する.まず,全変動を二次元画像に適用して離散化する際に,平方根計算が必要ないノルムを用いる.次に,劣こう配を用いた反復処理で骨格成分を計算する.双対変数を計算しないこと,FFTを含む射影関数や平方根計算が不必要,浮動小数点の計算回数の削減により計算量を削減する.計算時間を従来手法の(1)に対して53%,(2)に対して5%へと大幅に削減できることを実験結果により示す.提案手法は従来手法と同等の性能を維持したまま,陽に忠実化項と平滑化項から構成される反復処理を可能とする.

Divisible Load Theory (DLT) is an established mathematical framework to study Divisible Load Scheduling (DLS). However, traditional DLT does not address the scheduling of results back to source (i.e., result collection), nor does it comprehensively deal with system heterogeneity. In this paper, the DLSRCHETS (DLS with Result Collection on HET-erogeneous Systems) problem is addressed. The few papers to date that have dealt with DLSRCHETS, proposed Simplistic LIFO (Last In, First Out) and FIFO (First In, First Out) type of schedules as solutions to DLSRCHETS. In this paper, a new polynomial time heuristic algorithm, SPORT (System Parameters based Optimized Result Transfer), is proposed as a solution to the DLSRCHETS problem. With the help of simulations, it is proved that the performance of spoRT is significantly better than existing algorithms. The other major contributions of this paper include, for the first time ever, (a) the derivation of the condition to identify the presence of idle time in a FIFO schedule for two processors, (b) the identification of the limiting condition for the optimality Of FIFO and LIFO schedules for two processors, and (c) the introduction of the concept of equivalent processor in DLS for heterogeneous systems with result collection.

多重記述符号化の冗長度制御方式としてフレーム展開を用いる手法を提案する.ウェーブレット変換後の係数群をラティスにより複数のパケットに均等分割する従来手法では,パケット間の冗長度がラティスの構造に依存しており信号自体の冗長度には着目していない.そこで本検討では,信号を直接的に冗長表現するためにウェーブレット係数群をフレーム展開する.従来方式と比較して,パケットロスが生じる場合のエラー耐性が高まることを実験で確認した.

Divisible Load Theory (DLT) is an established framework to study Divisible Load Scheduling (DLS). Traditional DLT ignores the result collection phase, and specifies no solution to the general case where both the network speed and computing capacity of the nodes are heterogeneous. In this paper, the DLS with Result Collection on HETerogeneous Systems (DLSRCHETS) problem is formulated as a linear program and analyzed. The papers to date that have dealt with result collection, proposed Simplistic LIFO (Last In, First Out) and FIFO (First In, First Out) type of schedules as solutions. The main contributions of this paper are: (a) A proof of the Allocation Precedence Condition, which is inconsequential in LIFO or FIFO, but is important in a general schedule. (b) A proof of the Idle Time Theorem, which states that irrespective of whether load is allocated to all available processors, in the optimal solution to the DLSRCHETS problem, at the most one processor that is allocated load has idle time, and that the idle time exists only when the result collection begins immediately after the completion of load distribution.

Divisible Load Theory (DLT) is an established mathematical framework to study Divisible Load Scheduling (DLS). However traditional DLT does not comprehensively deal with the scheduling of results back to source (i.e., result collection) on heterogeneous systems. In this paper, the DLSRCHETS (DLS with Result Collection on HETerogeneous Systems) problem is addressed. The few papers to date that have dealt with DLSRCHETS, proposed simplistic LIFO (Last In, First Out) and FIFO (First In, First Out) type of schedules as solutions to DLSRCHETS. In this paper, a new heuristic algorithm, ITERLP, is proposed as a solution to the DLSRCHETS problem. With the help of simulations, it is proved that the performance of ITERLP is significantly better than existing algorithms.

2分割を繰り返すことによりコミック画像を各コマに分割する手法が提案されている.本論文では,幅をもつ検査帯により分割線候補の検出を行い,分割線適合検査によりコマ分割を決定する手法を提案し,実験によりしきい値と分割精度について評価を行った.

画像入力装置の普及により，容易に紙面のディジタル化が可能になった．さらに，画像入力時間の短縮のために解像度は低い方が良い．一方で，解像度の異なる様々な端末に画像を表示するためには解像度変換が必須である．我々はこれまでに，解像度変換が容易なベクター表現に着目し，ベクター変換を考慮した小数画素精度の2値化を提案してきた．しかし，線に振動が生じてベクター変換効率を低下させる場合がある．そこで，本稿では，均等色領域に隣接するエッジに対して，その方向性を考慮した1次元フィルタによる補間を行い，2値化により境界線を取得する手法を提案する．1次元フィルタには解像度変換において線の振動を低減するためにステップ関数を用いる．Digitizing paper have been convenient since image input devices are disseminated. Further, an input resolution is lower, an input speed is higher. On the other hand, a resolution conversion is mandatory technique in order to display the image on any panels which have varying resolutions. We have proposed subpixel binarization for vectorization, which is suitable for resolution conversion. However, vibrations of line drawings or border lines may occur, and then vectorization efficiency decreases. In this paper, we propose subpixel binarization based on a directive one-dimensional filtering along a border region of homogeneous tone. A step function is adopted as the one-dimensional filter for suppression of the vibration.

The major advantage of vector representation of an image is that the image quality is maintained for arbitrary scaling. In recent years, a demand for scalable image coding has been increasing because of the wide variety of available digital contents and display terminals. Conventional scalable coding schemes are based on raster representation, and thus, line drawings deteriorate in quality when expanded and shrunk. In this paper, we propose an edge reconstruction method using vector representation for the purpose of keeping a consistent spatial scalability on transmission and display. We take an anti-aliasing into account in edge areas for approximation of luminance values around the edge. The proposed method can improve PSNR by up to 2 dB as compared to the conventional methods when image is expanded and shrunk.

Vector representation of binary images has an advantage of keeping high image quality for arbitrary scaling as well as editing capability of an object. However, the vector representation suffers from low compression efficiency compared with JBIG. In this paper, we show the main cause reducing coding efficiency and propose two methods to improve it. The proposed methods can reduce the file size of a binary image up to about 30-40 percent.

H.264/AVCは符号化処理量が多い問題点がある. 多くはInter予測に関する処理であるが, 2種類の予測サイズが利用できるIntra予測も10%以上を占める. そこで異なる予測サイズのIntra予測結果を利用した処理量削減手法を提案する.

H.246/MPEG-4 AVCは高い符号化効率を実現する動画像符号化方式であるが、符号化処理量が多い問題点がある。符号化処理量の多くはInter予測に関連する処理が占めており、H.264/MPEG-4 AVCでは特にその比率が高い。従来の動画像符号化方式もInter予測の占める割合は高かったが、H.264/MPEG-4 AVCでは可変ブロックサイズ動き補償、2枚以上の複数参照フレームに加え1/4画素精度の動き補償が利用できるようになっており、符号化処理量増加の主たる原因となっている。そこで本稿では符号化対象マクロブロックと探索中に参照するマクロブロックの画素値の差分を利用することで動き補償ブロックサイズの探索を効率よく削減する手法を提案する。The performance improvement in H.264/MPEG-4 AVC is provided at the expense of higher computational complexity. Most of the complexity is caused by Inter prediction. To improve coding efficiency,some functions are added in H.264/MPEG-4 AVC,such as variable block size motion compensation,multi reference frame and quarter-pel motion compensation. A fast macroblock partition decision method is proposed in this paper. The macroblock size is efficiently determined by using the pixel value difference between encoding and the referred macroblock.

In this paper, we propose a method to vectorize comic images including halftone dots. Our method can prevent the jaggy and moire phenomena when the images are enlarged and shrank. The method contains three modules: halftone dots separation, gradation approximation, and vectorization. At the first module, small isolated areas and altered areas by dilation and erosion operations are separated into halftone dots images. At the second module, areas of halftone are approximated by contours and gradation parameters. At the last module, both halftone areas and line drawings are vectorized and approximated by smooth contours. The size of compressed file produced by our method is equal or smaller than JBIG compression. Validity of the proposed method is confirmed by experimental results.

著者らは，MPEG-4 Main Profileにおけるスプライトという新しい符号化ツールに着目し，動画像におけるカメラ操作のみを反映した背景領域にスプライトを適用する方法を検討してきた．スプライトが効果的な場合は，従来符号化法に比べて劇的な符号量の削減をもたらすことを明らかにしてきた．一方で，スプライトがどのような画像にも効果があるとは限らないということも知られており，適応的な利用が望まれていた．本稿では，スプライトを適応的に符号化へとりこみ，MPEG-4 Main Profile準拠の符号化システムとして開発した．本符号化システムは，スプライトが効果があると推定されるショットを自動的に抽出するスプライトショット抽出法を取り入れ，符号化品質の向上を図った．また，ストリーミング配信への適用も考慮し，低遅延スプライトに適用する2層ビデオオブジェクトレート制御法を新たに考案した．これにより，コンスタントビットレートの符号化データの生成を可能とした．また，カメラの動きのあるスポーツ映像を用いて，本提案符号化システムの評価を行い，その優位性と適応条件について考察した．<br>

我々は網点を階調表現に変換し，線画と共にベクトル表現に変換するマンガ符号化を提案してきた．階調近似処理において，従来手法では，固定サイズの小領域を複数個用いてグラデーションの向き，勾配，濃度を決定してきたが，モアレの発生により正しく濃度を決定できない場合があった．また，入力解像度や網点の線数の変化に対するロバスト性が欠けていた．本稿では，2値画像において網点周期を検出する手法を検討し，網点周期を利用してモアレの発生を低減させた濃度勾配検出法を提案する．シミュレーション実験により，提案手法の有効性を確認した．Comic coding by gradation approximation and vector representation has been proposed. When gradation approximation is processed in conventional method, we used some small areas with constant size. However, in some cases, we could not detect correctly because of the moire. In this paper, we study a detection method using a period of halftone dots. Then, we propose an improved method for density-gradient detection. Our approach decreases occurrence of moire. Through some simulation experiments, we have confirmed that our proposed method is valid.

Vector representation of graphics has an advantage of being displayed at any size. When resolution of a bitmap image is changed, lack of a line segment arises. In addition, moire occurs when resolution of an image with halftone dots is resized. In this paper, we propose a new technique to convert the binary image with halftone dots into the vector representation. Resolution conversion of binary image can easily be performed without moire by continuous tone approximation of halftone dots. First, we separate the area of halftone dots and line drawings in the image. Next, a continuous tone approximation is applied to the area of halftone dots. Then, the conventional vectorization is applied to both continuous tone areas and line drawings. Finally these components are mixed and reconstructed. Our approach provides an efficient way of displaying cartoon like images at any size with limited amount of data.

マンガは線画を中心に構成されているため，ベクトル化が有効な符号化手法である．しかし，単純なベクトル化は，網点の存在により逆に符号量の増加をまねく．また，網点を含んだままの解像度変換はモアレの発生原因となる．そこで，本稿では，マンガをベクトル表現する際に網点部分を分離して多階調近似し，残りの線分と共に符号化する手法について検討する．Vectorization is an effective technique for comic image coding, since comics are mainly consist of line drawings. However, an existence of halftone-dots causes an increase of coding bitrate if simple vectorization is used. In addition, moires occur when a resolution of the image with halftone-dots is changed. In this paper, we propose a new technique to achieve highly efficient comic image coding. First, we separate the area of halftone-dots and line drawings from an image. Then, a continuous tone approximation is applied to the area of halftone-dots. Next, the conventional vectorization is applied to line drawings. Finally, these two components are mixed together.

デジタルシネマやHDTVなどの超高解像度映像に対する符号化方式としてMotion JPEG2000の期待が高まっている．Motion JPEG2000ではwavelet変換を用いているため，解像度スケーラビリティが容易に実現できるといわれている．しかし，実際に用いられているwaveletフィルタは，現在解像度変換に用いられているデシメーションフィルタと比べると性能が悪く，また任意サイズに対応した解像度変換が実現できないといった問題がある．本研究では，最適解像度変換法の提案を目的とし，フィルタの性能比較や画質比較などの基礎検討を行った．Motion JPEG2000 is regarded as a promising coding system for an ultra high quality image like digital-cinema and HDTV. Wavelet transform used in Motion JPEG2000 enables to realize spatial scalability easily. However, the performance of the wavelet filter used in Motion JPEG2000 is worse than those of decimation filters used in image resampling and resizing. What is worse, the wavelet filter can only resize images to half resolution. We aim to provide of the best resampling method for Motion JPEG2000. In this paper, we investigate the performance of filters and its relationship with image quality.

The new MPEG-4 image coding standard that has recently been established offers higher image quality in low-bit-rate coding than conventional coding methods such as H.26X, MPEG-1, and MPEG-2, as well as coding for each object unit (a new function).
This paper proposes an algorithm for the automatic extraction of video objects consisting of a foreground object and a background sprite. It is based on "sprite coding," which can greatly reduce the number of code bits compared to conventional methods. The foreground and background video objects processed by the proposed method are subjected to MPEG-4 object coding and sprite coding, respectively, and are compared with MPEG-4 simple profile (normal) coding. In particular, the influence of the frame rate and the foreground object rate on the number of code bits was investigated. It was found that the proposed method can reproduce images with the same quality while using about one-fourth to one-half as many code bits as the normal coding method when the foreground rate is 10 to 15% of the whole image, regardless of the frame rate. In addition, the proposed method can attain more than double the frame rate of the normal coding method at a very low bit rate (e.g., 128 and 64 kbit/s) with a similar SNR. (C) 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 2, 86(6): 83-93, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjb.10158.

The new MPEG-4 image coding standard that has recently been established offers higher image quality in low-bit-rate coding than conventional coding methods such as H.26X, MPEG-1, and MPEG-2, as well as coding for each object unit (a new function).
This paper proposes an algorithm for the automatic extraction of video objects consisting of a foreground object and a background sprite. It is based on "sprite coding," which can greatly reduce the number of code bits compared to conventional methods. The foreground and background video objects processed by the proposed method are subjected to MPEG-4 object coding and sprite coding, respectively, and are compared with MPEG-4 simple profile (normal) coding. In particular, the influence of the frame rate and the foreground object rate on the number of code bits was investigated. It was found that the proposed method can reproduce images with the same quality while using about one-fourth to one-half as many code bits as the normal coding method when the foreground rate is 10 to 15% of the whole image, regardless of the frame rate. In addition, the proposed method can attain more than double the frame rate of the normal coding method at a very low bit rate (e.g., 128 and 64 kbit/s) with a similar SNR. (C) 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 2, 86(6): 83-93, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjb.10158.

Contents Delivery/Distribution Network (CDN) and Peer-to-Peer (P2P) network have been proposed for a large-scale contents distribution. We have been investigating what mechanisms are required for CDN, P2P, and both of them. It turns out that 1) optimal server selection and 2) content discovery are key role functions. In this paper, we propose a method for optimal server selection. The optimal server is defined to minimize end-to-end delay between a requesting host and the server. Here, end-to-end delay is regarded as network distance. Round trip time (RTT) can be used as a metric to predict end-to-end delay. However, it takes much cost to send a probe packet to a whole set of hosts in order to measure all RTT. Our proposed method send a probing packet to a set of servers called Landmarks, and predict which server is optimal based on these RTT. Simulation result shows the validity of the proposed method. We confirm that this approach can adapt to scalable architecture.

Contents Delivery/Distribution Network (CDN) and Peer-to-Peer (P2P) network have been proposed for a large-scale contents distribution. We have been investigating what mechanisms are required for CDN, P2P, and both of them. It turns out that 1) optimal server selection and 2) content discovery are key role functions. In this paper, we propose a method for optimal server selection. The optimal server is defined to minimize end-to-end delay between a requesting host and the server. Here, end-to-end delay is regarded as network distance. Round trip time (RTT) can be used as a metric to predict end-to-end delay. However, it takes much cost to send a probe packet to a whole set of hosts in order to measure all RTT. Our proposed method send a probing packet to a set of servers called Landmarks, and predict which server is optimal based on these RTT. Simulation result shows the validity of the proposed method. We confirm that this approach can adapt to scalable architecture.

著者らは，MPEG-4 Main Profileにおけるスプライトという新しい符号化ツールに着目し，動画像におけるカメラ操作のみを反映した背景領域にスプライトを適用する方法を検討してきた．スプライトが効果的な場合は，従来符号化法に比べて劇的な符号量の削減をもたらすことを明らかにしてきた．一方で，スプライトがどのような画像にも効果があるとは限らないということも知られており，適応的な利用が望まれていた．本稿では，スプライトを適応的に符号化へとりこみ，MPEG-4 Main Profile準拠の符号化システムとして開発した．本符号化システムは，スプライトが効果があると推定されるショットを自動的に抽出するスプライトショット抽出法を取り入れ，符号化品質の向上を図った．また，ストリーミング配信への適用も考慮し，低遅延スプライトに適用する2層ビデオオブジェクトレート制御法を新たに考案した．これにより，コンスタントビットレートの符号化データの生成を可能とした．また，カメラの動きのあるスポーツ映像を用いて，本提案符号化システムの評価を行い，その優位性と適応条件について考察した．<br>

著者らは，MPEG-4 Main Profileにおけるスプライトという新しい符号化ツールに着目し，動画像におけるカメラ操作のみを反映した背景領域にスプライトを適用する方法を検討してきた．スプライトが効果的な場合は，従来符号化法に比べて劇的な符号量の削減をもたらすことを明らかにしてきた．一方で，スプライトがどのような画像にも効果があるとは限らないということも知られており，適応的な利用が望まれていた．本稿では，スプライトを適応的に符号化へとりこみ，MPEG-4 Main Profile準拠の符号化システムとして開発した．本符号化システムは，スプライトが効果があると推定されるショットを自動的に抽出するスプライトショット抽出法を取り入れ，符号化品質の向上を図った．また，ストリーミング配信への適用も考慮し，低遅延スプライトに適用する2層ビデオオブジェクトレート制御法を新たに考案した．これにより，コンスタントビットレートの符号化データの生成を可能とした．また，カメラの動きのあるスポーツ映像を用いて，本提案符号化システムの評価を行い，その優位性と適応条件について考察した．<br>

MPEG-4では従来符号化法にはない，強力な符号化ツールが具備されている．筆者らはスプライトという符号化ツールに着目し，動画像をあらかじめ前景背景に分離し，前景にオブジェクト符号化，背景をスプライト符号化するスプライトモード[2]の開発を行ってきた．しかし，すべての動画像にスプライトモードが適応できるとは限らなかった．筆者らは，スプライト符号化を用いた汎用的なMPEG-4動画像符号化システムである"VideoESPER (Efficient SPrite aided EncodeR)"を開発した．スプライトモードが優位なショットを自動的に抽出し，そのショットにはスプライトモードを適用し，そのほかのショットに対しては従来符号化法を踏襲するノーマルモードを適用するマルチモード符号化法を提案する．さらに，AVIデータ等を逐次的に処理するための逐次処理方法，グローバルな動きの算出法，前景領域抽出法，スプライト生成法に改良を行った．また，符号化においては，前景背景に自動的に符号量を配分するためのマルチビデオオブジェクト(MVO: Multi Video Object)レート制御法を提案し，符号化効率を上げるために，スプライトに幾何変換を行う中間解像度変換法，スプライト回転法を提案する．また，実際に符号化実験を行って，従来符号化方式に対するVideoESPERの優位性を示した．一方で，VideoESPERにおける限界についても述べる．MPEG-4, a new audio-visual coding standard has strong compression tools, such as sprite coding. We have focused on sprite coding and proposed "Sprite Mode". In sprite mode, video sequence is divided into foreground object and background object. Foreground object is compressed using MPEG-4 object coding., while background object is compressed using sprite coding. Sprite mode does not always adopt any video sequence such that images having still camera motion can be compressed in enough quality in low bit-rate. So, in this paper, we developed new MPEG-4 video encoder using sprite coding , VideoESPER , which means Video Efficient Sprite aided EncodeR. In Video ESPER, multi-mode coding is newly proposed. Video shot suitable for conventional coding method is coded by normal mode (conventional method), while video shot suitable for sprite coding is coded by sprite mode. These two modes are automatic selected in any video sequences. We improved the processing flow using stuck system in order to consequently process video such as AVI format. We improved global motion detection algorithm, foreground object and background sprite generation algorithm. Then we proposed the simple rate-control for sprite mode and geometric sprite transforming method for more efficient sprite coding. Then, we have some experiment on MPEG-4 video coding. The result shows effectiveness' of VideoESPER in low bit-rate coding. On the other hand, limitation of ViddeoESPER is also discussed.

セル画アニメーション内には多数の均等色領域が存在する．この領域の輪郭線を有意点及び均等色を代表する色で近似することで圧縮率改善が図られる．そこで，本稿では輪郭線近似のための有意点をdynamic programmingを基に抽出し，輪郭を近似するアルゴリズムを提案する．Dynamic programmingを用いることで，誤差関数を基に，近似関数に最適な有意点を繰り返し抽出可能となる．Dynamicprogrammingは繰り返し演算に伴って，計算時間が大きくなることが問題である．そこで，本アルゴリズムの実行時間評価を行い，計算時間に影響を与える要素について検討を行った．一方，本アルゴリズムは，1)点数削減率を正確に制御できない，2)閾値によって輪郭線内に局所的に抽出点が発散してしまう問題がある．これらの問題を解決するためにFeedback algorithmを提案する．There are many homogeneous color regions in animation images. By approximating the regions by significant points and representative colors, coding efficiency can be improved. In this paper, we propose an approximation method for the regions based on dynamic programming. An approximation criterion used in the method is an error value defined by the distance between an original and an approximated line. If the error value is smaller than a threshold, a significant point that provides the minimum error value is extracted from the line. This process is iteratively executed until the error value exceeds the threshold. Since the the number of iteration tends to be large, we evaluate the processing time of significant elements, which is dominant for computational complexity. The remaining problems, in the proposed approximation, are 1) lack of point reduction ratio contorol and 2) excess extraction of significant points at some threshold. We also propose feedback algorithm in order to solve the problems.

In MPEG-4, a video sequence can be divided into foreground object and background objects that are independently encoded. Using sprites can dramatically compress the overall bit rate, but not all video sequences can be so encoded. This paper introduces MPEG-4 multimode coding
it offers automatic coding mode decision, video object generation and high compression efficiency. The source video sequence is segmented and each segment is automatically categorized as either "normal", which is encoded using MPEG-4 simple profile, or as "sprite". Coding experiments show that if the bit rate is low, multimode coding offers higher coding efficiency than regular MPEG-4 in terms of frame rate and image quality.

In MPEG-4, a video sequence can be divided into foreground object and background objects that are independently encoded. Using sprites can dramatically compress the overall bit rate, but not all video sequences can be so encoded. This paper introduces MPEG-4 multimode coding
it offers automatic coding mode decision, video object generation and high compression efficiency. The source video sequence is segmented and each segment is automatically categorized as either "normal", which is encoded using MPEG-4 simple profile, or as "sprite". Coding experiments show that if the bit rate is low, multimode coding offers higher coding efficiency than regular MPEG-4 in terms of frame rate and image quality.

In this paper, a method to extract moving objects on a pixel base for MPEG-4 "sprite coding" is proposed. Sprite coding is a form of object coding: it uses a unified panoramic background image derived from a sequence having a camera motion, and a foreground object as video object planes (VOP's). The proposed algorithm utilizes background difference, which is the difference between the original image and the background image, and watershed transform to extract the foreground moving objects. We first apply background difference to generate a foreground candidate image. Then, watershed transform is applied to this candidate image to extract the contours of the foreground objects. Furthermore, we utilize the macroblock mask of the foreground objects, which is generated by the conventional method, to delete the unnecessary edges extracted by watershed transformation. Results given by our proposed method are more than twice as better than that of the conventional method.

In this paper, a method to extract moving objects on a pixel base for MPEG-4 "sprite coding" is proposed. Sprite coding is a form of object coding: it uses a unified panoramic background image derived from a sequence having a camera motion, and a foreground object as video object planes (VOP's). The proposed algorithm utilizes background difference, which is the difference between the original image and the background image, and watershed transform to extract the foreground moving objects. We first apply background difference to generate a foreground candidate image. Then, watershed transform is applied to this candidate image to extract the contours of the foreground objects. Furthermore, we utilize the macroblock mask of the foreground objects, which is generated by the conventional method, to delete the unnecessary edges extracted by watershed transformation. Results given by our proposed method are more than twice as better than that of the conventional method.

In this paper, a method to extract moving objects on a pixel base for MPEG-4 "sprite coding" is proposed. Sprite coding is a form of object coding: it uses a unified panoramic background image derived from a sequence having a camera motion, and a foreground object as video object planes (VOP's). The proposed algorithm utilizes background difference, which is the difference between the original image and the background image, and watershed transform to extract the foreground moving objects. We first apply background difference to generate a foreground candidate image. Then, watershed transform is applied to this candidate image to extract the contours of the foreground objects. Furthermore, we utilize the macroblock mask of the foreground objects, which is generated by the conventional method, to delete the unnecessary edges extracted by watershed transformation. Results given by our proposed method are more than twice as better than that of the conventional method.

In MPEG-4, a video sequence can be divided into foreground object and background objects that are independently encoded. Using sprites can dramatically compress the overall bit rate but not all video sequences can be so encoded. This paper introduces MPEG-4 multi-mode coding; it offers automatic coding mode decision, video object generation and high compression efficiency. The source video sequence is segmented and each segment is automatically categorized as either "normal", which is encoded using MPEG-4 simple profile, or as "sprite". Coding experiments shows that if the bit rate is low, multi-mode coding offers more high coding efficiency than regular MPEG-4 in terms of frame rate and image quality.

A coding scheme specifically designed for animation images is proposed. Taking characteristics of animation images into account, lines and homogeneous color regions are extracted from animation images. Lines and contours of the homogeneous color regions are approximated by straightline and spline functions. We found that smoothing operations are effective to extract homogeneous regions from background images. Connected Filter is used for smoothing operation while keeping edges of the original image correctly. Precise approximation of the contours is also required. For this purpose, we propose to use Dynamic Programming with a feedback algorithm. As a result, an animation image can be represented by two layers; significant points as a base layer, and DCT blocks for an additional layer. For compensation scheme to the loss caused by smoothing operation, DCT is applied to the difference between the original images and lines and homogeneous color regions images. High-quality images can be obtained by adding this differential data to the approximated images.

In MPEG-4, a video sequence can be divided into foreground object and background objects that are independently encoded. Using sprites can dramatically compress the overall bit rate but not all video sequences can be so encoded. This paper introduces MPEG-4 multi-mode coding; it offers automatic coding mode decision, video object generation and high compression efficiency. The source video sequence is segmented and each segment is automatically categorized as either "normal", which is encoded using MPEG-4 simple profile, or as "sprite". Coding experiments shows that if the bit rate is low, multi-mode coding offers more high coding efficiency than regular MPEG-4 in terms of frame rate and image quality.

A coding scheme specifically designed for animation images is proposed. Taking characteristics of animation images into account, lines and homogeneous color regions are extracted from animation images. Lines and contours of the homogeneous color regions are approximated by straightline and spline functions. We found that smoothing operations are effective to extract homogeneous regions from background images. Connected Filter is used for smoothing operation while keeping edges of the original image correctly. Precise approximation of the contours is also required. For this purpose, we propose to use Dynamic Programming with a feedback algorithm. As a result, an animation image can be represented by two layers; significant points as a base layer, and DCT blocks for an additional layer. For compensation scheme to the loss caused by smoothing operation, DCT is applied to the difference between the original images and lines and homogeneous color regions images. High-quality images can be obtained by adding this differential data to the approximated images.

In this paper, a new system architecture called decoder-downloadable system is described. The purpose of this system is to provide a uniform platform for the multimedia world based on image compressions using characteristics of the contents. This system enables dynamic decoder downloading by negotiation between servers and clients for seamless and minimum delay playback. This negotiation scheme is implemented by Java and CORBA (Common Object Request Broker). Thus, servers and clients in this system are independent from OSes and hardware specifications. This system is suitable for TV broadcasting as well as Internet streaming.
We improve a decoder architecture in order to make the system more flexible. A decoder has some functions such as bitstream parsing, image compression algorithms (DCT, Wavelet, Motion estimation and so on). However, conventional decoders are monolithic software. Thus, it is impossible to share parts of decoders. The proposed method enables a module-based decoder and sharing some modules among some decoders. Consequently, the flexible decoder can make downloading time shorter by avoidance to download redundant parts of decoders. It also achieves scalabillity enabling this system to be used in some multimedia applications, such as a multimedia content search system, as well as a multimedia player.

In this paper, a new system architecture called decoder-downloadable system is described. The purpose of this system is to provide a uniform platform for the multimedia world based on image compressions using characteristics of the contents. This system enables dynamic decoder downloading by negotiation between servers and clients for seamless and minimum delay playback. This negotiation scheme is implemented by Java and CORBA (Common Object Request Broker). Thus, servers and clients in this system are independent from OSes and hardware specifications. This system is suitable for TV broadcasting as well as Internet streaming.
We improve a decoder architecture in order to make the system more flexible. A decoder has some functions such as bitstream parsing, image compression algorithms (DCT, Wavelet, Motion estimation and so on). However, conventional decoders are monolithic software. Thus, it is impossible to share parts of decoders. The proposed method enables a module-based decoder and sharing some modules among some decoders. Consequently, the flexible decoder can make downloading time shorter by avoidance to download redundant parts of decoders. It also achieves scalabillity enabling this system to be used in some multimedia applications, such as a multimedia content search system, as well as a multimedia player.

新しい画像符号化標準のMPEG-4が標準化されつつある。これは、低いレートにおいてはH.26XやMPEG-1,2に代表される従来符号化法と比較してより高品質の画像を提供するものである。また、新しい機能であるオブジェクト単位の符号化を提供するものである。筆者らはインターネット等に適用できる超低ビットレート符号化方法の開発を行っている。従来法に比べて劇的な符号量削減が可能であるMPEG-4の符号化ツールである「スプライト符号化」に着目し、前景、背景の2層からなるビデオオブジェクトの自動抽出アルゴリズムを提案する。生成した前景、背景のビデオオブジェクトをそれぞれMPEG-4オブジェクト符号化並びにスプライト符号化に適用(スプライトモード)して、スプライトを用いないMPEG-4 Simple profileの符号化法(ノーマルモード)と比較検討を行った。特に、フレームレートや前景オブジェクトの割合を変化させ、符号量に与える影響を調査した。その結果、フレームレートに関係なく、前景比率が画面全体の10〜15%程度である場合にMPEG-4通常符号化方法の1/4〜1/2程度の符号量で同程度の画質を実現できることを確認した。また超低レート(128kbit/s、64kbit/s)においてフレームレート、客観画質の評価を行ったところ、同程度のSNRで倍以上のフレームレートを達成できることを確認した。

インターネットでは, ユーザの接続環境により伝送レートや伝送経路の品質も様々である.このため, インターネット上で映像伝送サービスを提供するには, 符号化ビットストリームから, 実時間で任意の伝送レートに変換するビットストリームスケーリング技術が必要となる.従来, スケーリング方法として, DCT係数の符号量を削減滅する方法や, 他のフレームから参照されない符号化レームを間引く方法が提案されている.しかし, インターネット映像伝送サービスを提供するには, 従来方法のスケーリングの自由度では, 不十分である.そこで, 本論文では, 演算コストを抑えた参照フレームを間引くビットストリームスケーリング方法を提案する.提案方法は, 動きベクトル検索の代わりに, もとの動きベクトルから新たな動きベクトルを推定し, 動き補償予測の参照関係を変更する.これにより演算コストを抑えつつ, スケーリングの自由度が拡大できることを示す.また, 推定した動きベクトルと復号フレームから検出した動きベクトルを比較し, 動きベクトル推定方法の確度についても検証する.

MPEG-4 Main Profileでサポートされているスプライト符号化は，オブジェクト符号化方式の一種で，符号量を劇的に削減できる方式である。筆者らは，この符号化方式を利用し，インターネット等に適用できる高品質超低ビットレートを実現するアプリケーションの研究開発を行っている。しかし，スプライト符号化は全ての映像に適用できないという問題点がある。本稿では，スプライト符号化に適している映像を分析し，また，この判定を効率的に行う手法を提案する．Sprite coding is supported by the new video coding standard MPEG-4 Ver.1 Main Profile. It provides content-based functionality and low bit-rate video compression. We focused on this coding technique in order to develop high quality and very low bit-rate application that is effective in narrow-band transmission such as Internet. In this paper, we analyze that what kind of video sequence is suitable for sprite coding by checking a few parameters. This paper also proposes an automatic shot decision algorithm for sprite coding.

スタティックスプライトを用いた符号化では, スタティックスプライトの生成精度が符号化効率, SNR, 主観評価に大きく影響する.動画像からスタティックスプライトを生成するにはアウトライヤ(外れデータ)の影響のないカメラモーションだけを反映したグローバルモーションを算出する必要がある.本論文では, 局所的動きベクトルとその微分値の特徴空間での分布の特性を利用し, グローバルモーションの複数候補による検定手法を用いた, アウトライヤの影響の少ないグローバルモーション算出法を提案する.また, 提案手法を用いて算出されたグローバルモーションをMPEG-4VMVer.11に適用して, (1)LME符号化, (2)GME符号化, (3)ダイナミックスプライト符号化, (4)スタティックスプライト符号化を行い, それぞれの符号量と画品質を評価した.その結果, スタティックスプライト符号化はLME符号化, GME符号化, ダイナミックスプライト符号化に比べて高圧縮率を達成できることがわかった.

スタティックスプライトを用いた符号化では, スタティックスプライトの生成精度が符号化効率, SNR, 主観評価に大きく影響する.動画像からスタティックスプライトを生成するにはアウトライヤ(外れデータ)の影響のないカメラモーションだけを反映したグローバルモーションを算出する必要がある.本論文では, 局所的動きベクトルとその微分値の特徴空間での分布の特性を利用し, グローバルモーションの複数候補による検定手法を用いた, アウトライヤの影響の少ないグローバルモーション算出法を提案する.また, 提案手法を用いて算出されたグローバルモーションをMPEG-4VMVer.11に適用して, (1)LME符号化, (2)GME符号化, (3)ダイナミックスプライト符号化, (4)スタティックスプライト符号化を行い, それぞれの符号量と画品質を評価した.その結果, スタティックスプライト符号化はLME符号化, GME符号化, ダイナミックスプライト符号化に比べて高圧縮率を達成できることがわかった.

The new video coding standard MPEG-4 provides content-based functionality and low bit-rate video compression. We focused on the "sprite coding" supported MPEG-4 Version 1 Main profile in order to achieve "VHS quality video on 2B (128bps)" for narrow-band transmission such as the Internet. Automatic VOP (Video Object plane) generation technologies are being studied as one of the most important issues of MPEG-4 object coding. This paper proposes a two-layer VOP generation scheme with some core algorithms such as Gh IE (Global Motion Estimation). foreground moving object extraction, and background sprite generation. This paper also describes a shape information reduction method for foreground objects. Using this method, shape information is compressed by 90%. Experiments are conducted on VOP generation and video coding with MPEG-4. Coding efficiency is 3-4 times higher than that of typical existing video coding schemes at the same subjective image quality.

The new video coding standard MPEG-4 provides content-based functionality and low bit-rate video compression. We focused on the "sprite coding" supported MPEG-4 Version 1 Main profile in order to achieve "VHS quality video on 2B (128bps)" for narrow-band transmission such as the Internet. Automatic VOP (Video Object plane) generation technologies are being studied as one of the most important issues of MPEG-4 object coding. This paper proposes a two-layer VOP generation scheme with some core algorithms such as Gh IE (Global Motion Estimation). foreground moving object extraction, and background sprite generation. This paper also describes a shape information reduction method for foreground objects. Using this method, shape information is compressed by 90%. Experiments are conducted on VOP generation and video coding with MPEG-4. Coding efficiency is 3-4 times higher than that of typical existing video coding schemes at the same subjective image quality.

新しい画像符号化標準のMPEG-4が標準化されつつある.これは, 新しいファンクショナリティであるオブジェクト単位の符号化を提供するものである.著者らはインターネットアプリケーションなどに適用できる超低ビットレート符号化方法の研究を行っており, "VHS on 2B"(128kbpsでVHSビデオ並の品質)の符号化方式の開発を行っている.本稿では, MPEG-4 Version 1 Main profileにてサポートされている「スプライト符号化」に着目し, スプライト符号化の課題の一つであるビデオオブジェクトの自動抽出アルゴリズムを提案した.また, 提案アルゴリズムをMPEG-4符号化して従来符号化方法と比較検討を行った.同程度の画質を, 従来符号化方法の1/2以下の符号量で実現した.

PEG-LC is an international standard for multimedia coding that has been defined by ISO/IEC JTC 1/SC 29/WG 11. Conventional audio-visual coding is for compression of natural images, natural audio and speech, whereas, MPEG-LC is for compression of synthetic images such as computer graphics and synthetic audio as well as natural audio-visual sources. These sources are separately coded as different objects at the encoder so that they can be decoded according to user preference. This article describes MPEG-4 version one, for which the Final Draft International Standard (FDIS) voting phase has finished (1999).

画面全体に生じる動きや時間的輝度変化への対処法としてグローバル動き補償,グローバル輝度変化補償が提案されており,これらにより符号化効率の大幅な改善が期待できる.本論文では,特にグローバル輝度変化補償で問題であった膨大な演算量を低減するために,ブロック単位で行われる動き推定処理及び輝度変化パラメータ推定処理に簡易な手法を導入するとともに,その動き推定処理をグローバル動き補償,ローカル動き補償時の動き推定処理と共通化することにより,少ない演算量でこれらの機能を併せもつ符号化方式を提案する.実験で用いたブロックサイズ,動き探索範囲の場合,従来の各補償方法を単純に組み合わせる場合に比べ乗算回数が約1/400に低減される.また符号化効率の面では,グローバル動き補償と輝度変化補償を個別に適用した場合の効率改善効果を併せもつ効果が得られることを示す.

画面全体に生じる動きや時間的輝度変化への対処法としてグローバル動き補償,グローバル輝度変化補償が提案されており,これらにより符号化効率の大幅な改善が期待できる.本論文では,特にグローバル輝度変化補償で問題であった膨大な演算量を低減するために,ブロック単位で行われる動き推定処理及び輝度変化パラメータ推定処理に簡易な手法を導入するとともに,その動き推定処理をグローバル動き補償,ローカル動き補償時の動き推定処理と共通化することにより,少ない演算量でこれらの機能を併せもつ符号化方式を提案する.実験で用いたブロックサイズ,動き探索範囲の場合,従来の各補償方法を単純に組み合わせる場合に比べ乗算回数が約1/400に低減される.また符号化効率の面では,グローバル動き補償と輝度変化補償を個別に適用した場合の効率改善効果を併せもつ効果が得られることを示す.

画面全体に生じる動きや時間的輝度変化への対処法としてグローバル動き補償,グローバル輝度変化補償が提案されており,これらにより符号化効率の大幅な改善が期待できる.本論文では,特にグローバル輝度変化補償で問題であった膨大な演算量を低減するために,ブロック単位で行われる動き推定処理及び輝度変化パラメータ推定処理に簡易な手法を導入するとともに,その動き推定処理をグローバル動き補償,ローカル動き補償時の動き推定処理と共通化することにより,少ない演算量でこれらの機能を併せもつ符号化方式を提案する.実験で用いたブロックサイズ,動き探索範囲の場合,従来の各補償方法を単純に組み合わせる場合に比べ乗算回数が約1/400に低減される.また符号化効率の面では,グローバル動き補償と輝度変化補償を個別に適用した場合の効率改善効果を併せもつ効果が得られることを示す.

In this paper, a global brightness-variation compensation (GBC) scheme is proposed to improve the coding efficiency for video scenes that contain global brightness variations caused by fade in/out, camera-iris adjustment, flicker, illumination change, and so on. In this scheme, a set of tno brightness-variation parameters, which represent multiplier and offset components of the brightness variation in the whole frame, is estimated. The brightness variation is then compensated using the parameters. We also propose a method to apply the GBC scheme to component signals for video coding. Furthermore, a block-by-block ON/OFF control method for GBC is introduced to improve coding performance even for scenes including local variations in brightness caused by camera Bashes, spotlights, and the like, Simulation results show that the proposed CBC scheme with the ON/OFF control method improves the peak signal-to-noise ratio (PSNR) by 2-4 dB when the coded scenes contain brightness variations.

A hybrid coding algorithm, consisting of motion compensated interframe prediction and adaptive gain/shape vector quantization, is proposed for low bit rate video coding. A video coding system using the proposed coding algorithm and its coding characteristics are also described. Motion compensated interframe prediction is an efficient technique for reducing temporal redundancy contained in moving pictures. In the proposed scheme, constant amplitude blocks ("gray level blocks") are also included in search vectors in order to increase coding efficiency for quick object movement or scene change. Adaptive gain/ shape vector quantization with a tree search codebook is employed to encode motion compensated interframe difference signals. It can respond to changes in source statistics because its codebook is independent of the gain components of the input vectors. The proposed coding algorithm has been evaluated by computer simulations and shown to be effective for low bit rate video transmission. A 64 kbit/s video coding system based on this coding algorithm has been implemented. The developed coding system can transmit multiplexed video, audio and digital data at the basic ISDN rate, and can be applied to videophone as well as videoconferencing. © 1989.