Том 51, № 3 (2017)

Detection of abnormal video images of transportation is to find out video images that contain abnormities among all images of transportation using video and image processing and analyzing techniques. It is an important component of intelligent transportation system, which can not only reduce the workload of traffic managers, but also effectively improve the efficiency of traffic management. However, video images of transportation in practice usually have complex backgrounds, and current detecting algorithms of traffic abnormity sometimes become ineffective due to interference factors such as noises and affine transformation (illumination variation, target occlusion, scale changes and view changes, etc.). In order to overcome these interference factors and fuzzy uncertainties in image representation, as well as improve the accuracy of traffic images representation, this study explored the representation methods of traffic images using fuzzy geometry theory on the basis of fuzzy uncertainties occurring during the process of imaging, transmission and processing of images; moreover, this study also put forward two kinds of representation algorithms of traffic images, and analyzed and verified effectiveness of representation algorithms based on theories and experiments.

The paper discusses an algorithm for correcting code errors of the residue number system using a positional characteristic. Application of parallel-pipeline computations of this positional characteristic allows us to reduce hardware costs by 7.2% for processing 2-byte data represented in a residue number system code. The main properties of codes allow us to provide the required fault-tolerance for multi-rate DSP devices. The results provided in the paper can be applied to hydroacoustic monitoring tasks.

Pedestrian detection is a fundamental problem in video surveillance and has achieved great progress in recent years. However the performance of a generic pedestrian detector trained on some public datasets drops significantly when it is applied to some specific scenes due to the difference between source training samples and pedestrian samples in target scenes. We propose a novel transfer learning framework, which automatically transfers a generic detector to a scene-specific pedestrian detector without manually labeling training samples from target scenes. In our method, we get initial detected results and several cues are used to filter target templates whose labels we are sure about from the initial detected results. Gaussian mixture model (GMM) is used to get the motion areas in each video frame and some other target samples. The relevancy between target samples and target templates and the relevancy between source samples and target templates are estimated by sparse coding and later used to calculate the weights for source samples and target samples. Saliency detection is an essential work before the relevancy computing between source samples and target templates for eliminating interference of non-salient region. We demonstrate the effectiveness of our scene-specific detector on a public dataset, and compare with the generic detector. Detection rates improves significantly, and also it is comparable with the detector trained by a lot of manually labeled samples from the target scene.

As China’s economy develops vigorously, people’s living quality has been greatly improved and its car ownership has been increasing day by day. Consequently, traffic jams and safety problems become more and more severe. To solve these problems, the idea of car networking system construction has been put forward based on the improvement of Internet of Things and computer technologies. However, the vehicle terminals produced by many automobile manufacturers has disadvantages, such as incomplete functions, incompatible interface and exorbitant costs, etc. Therefore, based on the concept of car networking as well as the new 4G network technique, this study designed a kind of embedded car networking vehicle terminal and analyzed its hardware structure and installation problems. Moreover, its function of boot load program was extended and a software development platform of the embedded vehicle terminal based on Linux core was constructed. On the basis of the software and hardware environment construction, the global position system (GPS) and ZigBee wireless transmission software of the vehicle terminal were designed. In addition, the realization of the function of serial port communication was elaborated, and the development and realization of the embedded car networking vehicle terminal software system were explored.