Design Of Davinci-based Portable DVR System

With the continuous improvement of living, security issues gain people's widespread concern. As a safety precaution tool, video surveillance has been widely used in many aspects of people's life. Low cost, good stability and reliability, high integration, etc, all these advantages of the embedded DVR system make itself to be the major product in the video surveillance field. However, the most of the present DVR systems can only complete information acquisition, recording and replaying, but can't process and recognize the video information automatically. Currently, some embedded DVR systems with intelligent algorithms are developed by the companies, but the size is big and video surveillance algorithms are immature.This thesis proposed a portable intelligent DVR system. On the one hand, using the method of separating system into bottom board and core board, the system was made to be easy carry. On the other hand, by detecting moving target from collected video information automatically, the system was changed from passive monitoring to proactively identify, post-mortem analysis to pre-warning, and it could get the operator out of heavy monitor job. This thesis summarizes and analyzes the basis of relevant research at home and abroad, and expands the following research and development works:1.Hardware design and the concrete realization of each module of the portable intelligent DVR system which based on Davinci chip TMS320DM6446 processor and Linux operation system were proposed. Comprehensively considering signal integrity and power integrity issues, PCB layout was completed.2.A moving target detection algorithm was proposed based on Gaussion model. By using statistical method of histogram and foreground probabilities to compensate the single-Gaussion background model, moving target detection was achieved. And implemention of the algorithm on the hardware was carried out.3.Using the system software data flow analysis, embedded DVR software was divided into multiple task modules, including command interpreter, system settings and files query task, video capture and moving target detection task, video store task, decompression processing task, video play task. The concrete implementation of each task module and the inter-program communication were designed. The performance of the system had been tested and evaluated.Finally, the works were summarized. The aspects where need to be improved and the works followed were pointed out.