| With the rapid development of intelligent technology, the advent of 3G wireless communications and the internet age, accelerate to the application research of network intelligence video surveillance system. Video surveillance system has also seen from analog to digital, network-based and intelligent development process. Motion detection and tracking technology is an important part of computer vision, also, is the basis for realization of intelligent video surveillance system. Moving object detection and tracking technology has been a hot research topic in recent years because of its wide application prospect. At the same time, with the embedded technology getting mature gradually, a variety of high performance multimedia processors continue to emerge. There must be a wave of intelligent surveillance within digital video signal based on embedded system.This thesis is based on the above background. We analyzed several typical motion detection and tracking methods, then presented several corresponding improvement algorithms of motion detection and tracking. On this basis, by means of TMSC6000DM642 embedded development platform, which is manufactured by Texas Instrument Inc, we designed and realized two sets of (C / S and B / S architecture) network intelligent video surveillance system based on embedded system.The main contribution of this thesis and the achievement are as follows:1) In motion detection module, an improved and robust algorithm for motion detection from complicated background in video sequences is proposed. The algorithm is based on background edge-characteristic difference combining inter-frame difference and chromaticity distortion.2) In tracking objects module, the adaptive bandwidth MeanShift algorithm, which is based on weighted color histogram, is adopted to enhance the tracking robustness. To solve the occlusion problem in single object tracking, an algorithm of trajectory predictor which combine normalized Kalman forecast with simple linear predictor is designed. A synthesis algorithm based on motion estimation, object texture and target gray is designed to enhance the performance. Finally, the thesis realized multiple moving objects detection and tracking algorithm.3) In DSP module, two kinds of embedded intelligent video surveillance system based on TMS320DM642 embedded platform which is manufactured by Texas Instrument are proposed. By using DSP / BIOS operating system and RF-5 framework technology, two sets of intelligent monitoring system based on C / S and B / S architecture are set up. Thus the DSP-based intelligent video surveillance system is realized with the function of video capture, H.264&JPEG-based compression algorithm, intelligent algorithm, network application services, PTZ camera control and alarm based on GSM.4) In client module, a highly flexible client system based on windows system is implemented. Through the client system, the DSP based system is under control, including functions of PTZ camera control, video capture, network settings, multi-channel DSP display, intelligent algorithms on / off, database management et al. |
