Research And Design Of Moving Target Tracking System Based On Dual-core Embedded Platform

With the development of urban construction and social economy, especially the "smart city" and "intelligent transportation" projects are put forward, the field of security has brought the unprecedented development opportunity. At the same time, with the rapid development of large scale integrated electronic technology, hardware manufacturing technology and multimedia information technology, the data processing and analysis capabilities of processor has been greatly improved, the intelligent video analysis system based on embedded platform occupies more and more market because of its high level of integration, high speed of data analysis ability, strong system stability and high speed of data transfer, etc.This paper mainly focuses on the intelligent video analysis system based on the front end, where the video data is analyzed to get the interested information at the time of collecting data, while the back-end server only needs to store the interested information, so that it can effectively reduce the server’s pressure of data processing and storage. At the same time, when an emergency event is detected, the alarm signal can be sent immediately, then the staff can be notified to deal with it, without waiting for the server to analyze all of the video data. The DaVinci digital media processing platform TMS320DM6467T of TI company, has high-speed data processing capability, simple video processing algorithm interface, perfect image and video processing library, and can effectively reduce the cost of video processing algorithms transplanted to it, and can reduce the cost and time of intelligent video analysis system development.In the field of intelligent video surveillance, moving target detection and tracking algorithm is always the important and difficult point, and the related algorithms are continuously proposed or improved to increase the accuracy of detection and tracking. In the process of moving target tracking, at first, in order to separate the moving object and background, the detection algorithm is used to process video sequence of the scene. After selecting the moving object that need to track, the moving target tracking algorithm is used to extract the specific position of the moving object in video sequences, then the motion trajectory of moving targets is recorded for later retrieval and query, so that the staff can be freed from the onerous task of monitoring.Now most of the moving object tracking algorithms are very complex, it’s difficult to transplant it to DaVinci platform for real-time processing directly. If the target is moving, or the illumination changes obvious or the moving target stops suddenly, in Vibe algorithm calculation, the old moving object (the ghost) can’t disappear from the foreground image for a very long time, which will affect the accuracy of target tracking. In this situation, the Vibe algorithm is improved in this paper to eliminate the influence of the ghost. An improved moving target tracking algorithm is proposed in this thesis by combing the Mean Shift algorithm with Vibe algorithm, it is transplanted into DaVinci platform for simulation and experiment after optimization, the experiment result showed that the improved algorithm can efficiently and track moving targets in real time.In this paper, I also study the software framework and hardware circuit of TMS320DM6467T platform. At first, I analyze each functional module of hardware circuit, and give the design scheme of each module according to the needs of the system, including video acquisition module, network transmission module, data storage module, the serial communication module, etc. Finally, I analyze the software framework of the system, and realize the video image acquisition, encapsulation and transplantation of DSP algorithms, network transmission, and other functions of the software design.