Research Of Real-time Moving Object Detection And Tracking System Based On Video

Intelligent Video Surveillance System which based on Computer Vision technology hasalready been widely applied to various aspects of human’s production and life at present.Moving object detection and tacking is the key technology and primary research direction ofIntelligent Video Surveillance System. In order to real-time detecting moving object emergedfrom video sequence frames, obtaining object parameters, matching and tracking movingobject, then the movement locus of object is acquired. In this paper major researches includethe structure of moving object detection and tracking system based on video, the technologyof video and image processing used in system, the algorithm of moving object detection andtracking, and the implementation of system based on computer and embedded technology.Firstly, conventional algorithms of moving object detection and tracking including eachadvantage and defect are analyzed and researched, and a kind background subtractionalgorithm based on accumulated difference update is proposed. The algorithm incorporatedaccumulated difference notion improvements self-adaptive background modeling algorithm toeliminate noise interference. A kind partition method of gray histogram threshold is proposed.Then, CAMShift algorithm and Kalman filter algorithm are analyzed and compared, anda kind moving object tracking algorithm which combines Kalman filter based on geometriccharacteristic with object histogram matching is proposed. The center of mass of object’sgrayscale and the length and width of circumscribed rectangle frame are used as the geometriccharacteristic parameter of object. After using respectively two groups of Kalman filter tocalculate, matching object according to similarity of template of object’s grayscale.And then, the moving object detection and tracking system based on USB camera isimplemented by VC++6.0Integrated Development Environment. The MFC applicationprogram wizard and Intel OpenCV code are utilized to develop system. According to inputtedvideo from either USB camera or AVI files, the system can detect and track object in scene.Finally, the real-time moving object detection and tracking system based on SOPC isdesigned. The Altera DE2board which contains CycloneⅡseries EP2C35F672C6FPGA chipis used as development platform. The Developer Kits include QuartusⅡ, SOPC Builder andNiosⅡIDE etc. The hardware design and software programming are accomplished. Theembedded system is able to partly operate after simulating and debugging.