Research On And Development Of An Intelligent Video Surveillance System Based On DM642

With the increasing demand of social security,the technology of Intelligent Video Surveillance(IVS) is becoming one of new arising high-tech application fields.Its main goal is:to automatically realize the detecting,description,tracking,and analysis of the moving objects in the surveillance video by integrating computer vision technology, image/video processing technology and artificial intelligence technology;and then to understand the meanings of video stream and explain the scenes comprehensibly;hence to take corresponding measures to react for accidents that may occur.To the technology of IVS,there are still many problems both in theory research and in applications fields.A great amount of researchers devoted themselves into the field and have already achieved many progresses.The dissertation focused on the key technologies of IVS based on the previous works.The main research content of the paper is the moving object detection,tracking,and behavior analysis.In order to overcome the existing systems' shortcomings,such as based on PC,higher cost,less reliability and portability, and failure to satisfy real-time processing,this dissertation studied and accomplished the DSP implementation of IVS.Regarding moving object detecting(MOD),several object detection algorithms used generally were introduced.Then,a novel approach to detection of Video Moving Objects (VMOs) was proposed in this dissertation,which combined grads-based frame difference with background subtraction.The experiment results showed that the proposed approach has good effect of MOD.Regarding moving object tracking(MOT),considering the complexity and characteristics of video object tracking,an improved MOT algorithm was proposed by combining the Kalman tracking algorithm with the Mean-Shift algorithm.The experiment results showed that the proposed approach is feasible,effective,and robust. Regarding moving object behavior recognition(MOBR),several main methods of behavior analysis were introduced.Meanwhile,to satisfy the actual requirements,an approach of behavior analysis based on three-layer model was given in detail,which was improved by integrating the specified surveillance background in this paper.The improved method was competitive and effective by experiments.Finally,the characteristics and the software development of TI DM642 were studied, and the code of above algorithms was transplanted into DM642.A prototype system of IVS was developed.By optimizing the related code in the system,the operating speed was improved greatly,and the real-time requirement was satisfied.