Research And Implementation Of Video Surveillance System Based On Dm642

With the growing of Video encode/decode,and RTOS,there must be a wave of security surveillance within digital video signal based on embedded system.After the mixing stage of analogy and digital,video surveillance system will inevitably develop towards to digital,network and intelligent directions.Because of the limitation of network bandwidth,That how to keep higher rate of video compressing base on ensureing high quality of image is the key of the success of a network video surveillance system.This paper makes video compressing as research breakthrough point.It realizes the video compressing and the network transmission at the DM642 EVM platform,and makes certain exploration in the movement image division and movement goal detection at end.The main research works and results are as follows.⑴The network video supervisory system design is given at first.We analyze MPEG-4 encoder and decoder;mainly studying the sharp encoding,motion estimation,compensation and texture encoding. We research the mode of intra and inter,DC/AC coefficient prediction as well.⑵Analyzing MPEG-4 encoding and decoding deeply,then combining the hardware features of DM642 EVM,we use many methods to optimize code.Through the test of video sequence,we improve the speed of encoding and decoding as well as ensure higher rate of compressing.Finally,it meets the requirement of real-time encoding and decoding.⑶A visual surveillance system is constructed by combing DM642 EVM developing platform with PC through the Ethernet.Video coding,video decoding and network transmission are realized on the system.The testing results show that this network video monitor system is designed with high stability,high image quality,good network transmission performance,and it can meet the goal of research and development.⑷An improved edge detection algorithm has been worked out in image segementation and has been applied into motion detection.It adopts improved background difference method for motion detection.The test on hardware platform shows that the effects of motion detection are improved obviously by the algorithm.