For next-generation logistics systems using autonomous vehicles and drones, spoofing of the GNSS location data induces serious problems. Although signal-based anti-spoofing has been studied, it is difficult to apply to current commercial GNSS modules in many cases. We investigate possibilities to detect spoofing of GNSS location data using multiple sensing devices and a decision tree classifier. Multiple features using the GNSS, beacons, and the IMU are defined and create a model to detect spoofing. Experimental results using learning-based classifier indicates the higher performances and generalization capability. The results also show that distance from beacons is useful to detect GNSS spoofing and indicate prospects of installation for the future drone highways.

We are studying a method to detect suspicious objects carried by a person in his/her clothing using both passive and active imager images by W-band radar-based sensing technology. In order to improve the accuracy of suspicious object detection, in this paper, we propose a two-stage detection method using time-series images of a walking person, where first the presence or absence of suspicious objects is checked, then the identification of the object type is performed if high probability of suspicious object presence is detected in the first stage. According to the experiment, the proposed method is effective in terms of the detection accuracy compared with a method that only identifies the type of suspicious objects.

Distributed ledger technology is becoming popular these days because of its high confidentiality, decentralization, and nontampering. It is suitable for replacing centralized security disadvantaged point transfer systems. Low-power wide area network (LPWAN) is capable for long-range communication with low-power consumption. The iconic features like wide area coverage and long battery-powered duration make it best to combine with large-scale IoT application deployment. In both industry and academic field, such combination of LPWAN and point transfer system is highly attended. However, the ledger management system generates too much data that low-bandwidth network such as LPWAN can hardly handle; meanwhile, the processing power’s requirement for small IoT devices is challenging. Towards addressing these issues, we design a packet transmission optimizing mechanism for a ledger-based point transfer system (LPTS) in LPWAN to reduce overall data traffic and build a simulator to evaluate its performance. Moreover, we have implemented the system and evaluated in field experiment.

In recent years, terrorist attacks have been spreading worldwide and become a public hazard to human society. The suspicious object detection system is an effective way to prevent terrorist attacks in public places. However, traditional systems face two main challenges: First, they need to conduct security checks at the entrance one by one, which leads to crowding; second, they rely heavily on screeners’ ability to understand security images, which can easily lead to misjudgment. To address these issues, we propose an AI-based W-band suspicious object detection system for moving persons that can perform a two-stage walkthrough screening for suspicious objects in an open area to maintain high throughput. The 1st screening uses millimeter wave radar and cameras to automatically screen suspects who may have concealed suspicious objects in an open area. The 2nd screening involves security personnel using a hybrid imager with active and passive imaging capabilities to identify the specific suspicious objects carried by the suspect. Convolutional neural network (CNN) based artificial intelligence (AI) technology will be used to improve the accuracy and speed of suspicious object detection. We performed an experiment to validate the proposed system. The usability and safety of the system are demonstrated by recognition rate (aka accuracy rate) or both recall and precision rate. In addition, in the process of improving the suspicious object recognition rate by AI techniques, we use generative adversarial network to help build a suspicious object database and successfully validate the effectiveness of the method and the factors affecting the suspicious object recognition rate to optimize the system.

Low Power Wide Area Network (LPWAN) is a long-range low-power wireless communication network. Its features, such as wide network coverage and low power consumption of terminals, make it suitable for large-scale deployment of IoT applications. The points transfer system, especially points transfer system in LPWAN, as a typical third-party payment application, is being closely attended by both industry and academia. Recent studies have shown that distributed ledger technology, because ofits characteristics such as high confidentiality, non-tampering, and decentralization, is a good solution to problems such as low-security performance due to centralized storage for a points transfer system. However, the distributed ledger will generate a large amount of data traffic in recording the transactions of network participants, which is a challenge for resource-constrained IoT devices. To address these issues, we propose an optimized packet transmission mechanism for a ledger-based points transfer system in LPWAN. Simulation results show that our proposed mechanism can well reduce the packet transmission ofthe whole system and meet the requirements of LPWAN. Moreover, we update the reader with information about distributed ledger and standardization-related activities in this paper.

Low Power Wide Area Network (LPWAN) is an Internet of things (IoT) network layer technology that has emerged in recent years for long-range and low-power communication needs in IoT. Its low-bandwidth, low-power, long-range and mass-connected IoT application features can be well applied to the points transfer system. However, the traditional centralized points transfer system faces many problems such as centralization, high computational requirement of nodes, and low robustness which are difficult to be widely used. In order to solve these problems, we propose a distributed ledger-based points transfer system in LPWAN and analyze the system robustness from the disaster management aspect. The simulation results show that our proposed system can still have strong robustness under extreme disaster situations and ensure the safe and efficient operation of the whole system.

In application related to public security check system, passive and active imaging of millimeter wave still faces critical challenges in providing high resolution quality images. Improving the detection, localization, and recognition accuracy of concealed object detection systems is very challenging due to the lack of a dataset of millimeter wave images with good resolution. Although previous studies proposed artificial intelligence-based concealed object recognition systems, improving accuracy remains a critical challenge. Therefore, in this paper, we propose two kinds of training dataset generation methods based on the proposed active millimeter wave imaging (AMWI) approaches presented in our previous work to improve the accuracy of convolutional neural networks (CNN)-based concealed object recognition systems. First, a depth-based training dataset generation method and a distance-based training dataset generation method are proposed for specular images and nonspecular images. Finally, a CNN-based concealed object recognition system is proposed by using generated active millimeter wave images and interferometer active images to improve the recognition accuracy.

To reduce safety violations in wide-area ranges, there is a need for highly functional multicamera surveillance systems. We introduce a multicamera trajectory forecasting surveillance network system based on a content-oriented suspicious object network system. This system uses multiple cameras in detection and recognition to track persons among different areas and is capable of retracking people. Each camera node has a processing unit and uses information-centric networking technology to build a content-oriented IoT network. We use field-recorded data to support the simulation, and the evaluation result indicates that our trajectory forecasting method is more efficient than conventional surveillance systems.

Pedestrian position estimation in videos is an important technique for enhancing surveillance system applications. Although many studies estimate pedestrian positions by using human body detection, its usage is limited when the entire body expands outside of the field of view. Camera calibration is also important for realizing accurate position estimation. Most surveillance cameras are not adjusted, and it is necessary to establish a method for easy camera calibration after installation. In this paper, we propose an estimation method for pedestrian positions using face detection and anthropometric properties such as statistical face lengths. We also investigate a simple method for camera calibration that is suitable for actual uses. We evaluate the position estimation accuracy by using indoor surveillance videos.

Along with the rapid development of IoT technology, Low power wide area network (LPWAN) has become the primary technology for IoT access today due to its characteristics of low cost, low power consumption, long-distance, and mass connections. At the same time, the points transfer system, as a typical third-party payment application, is attracting more and more extensive attention from academia and industry. Therefore, the research and development of a points transfer system for LPWAN are of great practical importance. However, the current points transfer systems often face problems such as centralization, high requirements on node computing power, and low robustness, which are difficult to adapt to the development of IoT. To address these problems, we propose a lightweight ledger-based points transfer system for application-oriented LPWAN. The system enables the points transfer between tags in LPWAN, where both tags and nodes are limited. Furthermore, it can be utilized even in disaster situations. Experimental results show that our proposed system is intensely robust and has a lower processing time than Bitcoin-like systems to cope with LPWAN's requirement.

Artificial Intelligence (AI) based millimeter wave radiometric imaging has become popular in a wide range of public security check systems, such as concealed object detection and identification. However, the low radiometric temperature contrast between small objects and low sensitivity is restricted to some extent. In this paper, an advanced radiometric passive imaging simulation model is proposed to improve the radiometric temperature contrast. This model considers additional noise, such as blur, variation in sensors, noise sources and summation of the number of frames. We establish a comprehensive training dataset that considers the physical characteristics of concealed objects. It can effectively fill the lack of a large database to avoid deteriorating the identification accuracy of AI applications. Moreover, it is also a key solution for improving the robustness of AI based object identification by using a convolutional neural network (CNN). Finally, simulation results are presented and analyzed to validate the proposed comprehensive training dataset and simulation model. Consequently, the proposed simulation model can effectively improve the robustness and accuracy of AI-based concealed object identification.

Artificial Intelligence (AI) based millimeter wave radiometric imaging has become popular in a wide range of public security check systems, such as concealed object detection and identification. However, the low radiometric temperature contrast between small objects and low sensitivity is restricted to some extent. In this paper, an advanced radiometric passive imaging simulation model is proposed to improve the radiometric temperature contrast. This model considers additional noise, such as blur, variation in sensors, noise sources and summation of the number of frames. We establish a comprehensive training dataset that considers the physical characteristics of concealed objects. It can effectively fill the lack of a large database to avoid deteriorating the identification accuracy of AI applications. Moreover, it is also a key solution for improving the robustness of AI based object identification by using a convolutional neural network (CNN). Finally, simulation results are presented and analyzed to validate the proposed comprehensive training dataset and simulation model. Consequently, the proposed simulation model can effectively improve the robustness and accuracy of AI-based concealed object identification.

The threat of terrorism has spread all over the world, and the situation has become grave. Suspicious object detection in the Internet of Things (IoT) is an effective way to respond to global terrorist attacks. The traditional solution requires performing security checks one by one at the entrance of each gate, resulting in bottlenecks and crowding. In the IoT paradigm, it is necessary to be able to perform suspicious object detection on moving people. Artificial intelligence (AI) and millimeter-wave imaging are advanced technologies in the global security field. However, suspicious object detection for moving persons in the IoT, which requires the integration of many different imaging technologies, is still a challenge in both academia and industry. Furthermore, increasing the recognition rate of suspicious objects and controlling network congestion are two main issues for such a suspicious object detection system. In this paper, an AI-based W-band suspicious object detection system for moving persons in the IoT paradigm is designed and implemented. In this system, we establish a suspicious object database to support AI technology for improving the probability of identifying suspicious objects. Moreover, we propose an efficient transmission mechanism to reduce system network congestion since a massive amount of data will be generated by 4K cameras during real-time monitoring. The evaluation results indicate that the advantages and efficiency of the proposed scheme are significant.

We propose vehicle wheel detection and axle counting techniques using LIDARs (Laser Imaging Direction and Ranging) for toll collection systems. Commonly-available LIDARs have insufficient scan speeds to count axles for high-speed vehicles. We install two LIDARs at left and right sides of the road to compensate for low scanning resolution. Range data of LIDARs are transformed into orthogonal coordinates that represent the depth from the LIDARs and the height from the road surface. Wheel candidates are extracted by using depth histograms above the road surface. The wheel candidates of two LIDARs are integrated and verified to reduce impacts of noises and missing data of wheel candidates. Experimental results using 206 transit data explain that our approach makes axle counting robust and reduces counting errors.

In this paper, we propose a camera-based probe car system for traffic condition estimation. The probe car, which mounts a stereo camera and a GPS sensor, measures the traffic density using a car distance map from the stereo camera and receives the velocity and position information of the car from the GPS sensor. A traffic control center creates a traffic condition map according to the probe car data and uses it for traffic control, car navigation, and bus/taxi scheduling. We implement the camera-based probe car system as an in-vehicle device using an image recognition processor and show its enough performance and speed for practical use.

In this paper, a vehicle detection framework for road surveillance systems is presented. The framework is based on co-occurrence histograms of oriented gradients (CoHOGs), which are effective for object detection, using the ViscontiTM 2 high-performance image recognition processor. We have developed vehicle detection and tracking technologies on the platform and confirmed these technologies can be applied to road surveillance systems.

We propose accurate vehicle detection using stereo images for toll collection systems. Conventional methods such as inductive loop sensors and laser scanners have disadvantages for equipment and installation costs. We have developed stereo vision-based depth measuring methods to detect a body of vehicles that achieve higher detection rates. Configurations of stereo cameras and algorithms are specialized for toll collection systems. Traffic data over 10,000 vehicles are used to confirm the accuracy of detection. The results indicate the proposed our approach will realize sufficient performance for toll collection systems and realize low cost vehicle detection for intelligent infrastructures for transportation.

This paper presents a method to automatically align a pose of 3D shape data to fit another shape data taken from different viewpoints. One of the difficult issues is to handle shape data which have surface information in different sides due to the difference in viewpoints, and to deal with objects in different scale. We detect local feature points on the two shape data, make potentially corresponding pairs of three feature points, calculate transformation parameters to align the three points, and get optimal alignment parameters by the voting of parameters obtained from the pairs of three points. We used hash table to avoid combinatorial explosion in making the pairs, and used geometric invariants for its key which are calculated from the positions of the points to keep the scale invariance. The method was evaluated with some public data and a set of laser-scanned data, and proved to be effective in alignment of shape data in different angles or scales.

Various applications of license plate recognition are increasing rapidly, for uses at electronic toll collection gates, parking areas, etc. However, there are difficulties in accurate recognition using conventional low resolution ITV cameras that are used for widespread video monitoring systems. In this paper, a license plate recognition system is proposed for low resolution ITV images.

An access control system that automatically logs in a user by his or her facial information is described. Furthermore, a method for automatically tracking logged data, based on a repetitive operation of template improvements, is shown. Face recognition is a favorable variety of biometrics for personal identification because user's facial images can be captured successively from a standard video camera at a distance. Two models of the system using face recognition were tested to prevent fraud and to create automatic registration. Experimental results for a six-month test showed that the tracking capability of the method is quite practicable
the equal error rate for false rejection and false acceptance(EER) is only about one percent.

A face-recognition security system for access control is described that is unaffected by the stance of the entrant. Although face recognition has several advantages compared to other biometric techniques, the similarity value used for recognition is affected by the direction that the entrant is facing, the entrant's position relative to the equipment, and the lighting conditions. The proposed system eliminates the second problem by using feedback to stabilize the entrant's position and an efficient method to update the registered data. A six-month evaluation of a prototype system showed that it had a FAR(false acceptance rate) of 0.1% and a FRR(false rejection rate) of 1%. Simulation of one-to-many identification showed that contact-less security is feasible using this system.

The automatic extraction and recognition of news captions and annotations can be of great help locating topics of interest in digital news video libraries. To achieve this goal, we present a technique, called Video OCR (Optical Character Reader), which detects, extracts, and reads text areas in digital video data. In this paper, we address problems, describe the method by which Video OCR operates, and suggest applications for its use in digital news archives. To solve two problems of character recognition for videos, low-resolution characters and extremely complex backgrounds, we apply an interpolation filter, multiframe integration and character extraction filters. Character segmentation is performed by a recognition-based segmentation method, and intermediate character recognition results are used to improve the segmentation. We also include a method for locating text areas using text-like properties and the use of a language-based postprocessing technique to increase word recognition rates, The overall recognition results are satisfactory for use in news indexing. Performing Video OCR on news video and combining its results with other video understanding techniques will improve the overall understanding of the news video content.

Video OCR is a technique that can greatly help to locate topics of interest in a large digital news video archive via the automatic extraction and reading of captions and annotations. News captions generally provide vital search information about the video being presented, the names of people and places or descriptions of objects. In this paper, two difficult problems of character recognition for videos are addressed: low resolution characters and extremely complex backgrounds. We apply an interpolation filter, multi-frame integration and a combination of four filters to solve these problems. Segmenting characters is done by a recognition-based segmentation method and intermediate character recognition results are used to improve the segmentation. The overall recognition results are good enough for use in news indexing. Performing video OCR on news video and combining its results with other video understanding techniques will improve the overall understanding of the news video content.