Study On Detection And Tracking Algorithm For Single Target Motion In Intelligent Video Surveillance

Intelligent video surveillance is an important part of the field of computer vision, it has been widely used in the military, medical, security and scientific research and other fields. The design of detection and tracking algorithm is its most two basic technologies in the intelligent video Surveillance system. To study these two basic technologies is very meaningful to enhance the performance of intelligent video surveillance system. This paper focuses on the technology of moving target detection and tracking in the video, Mainly works are as follows:While studied in the field of Motion’s target detection algorithm,first we targeted on and compared three common motion’s target detection methods: backgroud difference method, optical flow method and frame dfiference method. Focuses on the five differential algorithm, Aimed at the problem of some holes and uncontinuous edges in the goals that detected by the five differential algorithm.Propose to use an improved algorithm based on five differential method to detect moving targets. Its principle is that by means of Extracting five continuous video frame image, take two out of five frame to make four frame differential, then use the results of this four times frame differential to do two-two OR operator, after that do And operator to get the moving target.Compared the detect effects of the tfive differential and improved algorithm basedon the five differenctial method through experiments. Experiments show that employ the improved algorithm based on five differenctial method to extract the moving target, it effectively eliminating the problems of voids and edge blur, more preferably results achieved.As to the research on Algorithm of moving object tracking, focuses on the analysis of the meanshift algorithm’s application in target tracking.Against that traditional meanshift tracking algorithm can not effectively eliminate background information contained within the target,and can not be adaptive to the significance changes of background in the continuous video.Proposed an improved target tracking algorithm based on meanshift. When calculating the model histogram of moving object and the background, the improved algorithm use the differential of their bin value of histogram to extract significant size of the target characteristic values, and embedded this value into the traditional measure of similarity coefficients by the form of the weight, using the moment information of the target to measure the intensity and angle changes of the target.Compared the simulation results of the improve meanshift tracking algorithm and classical meanshift algorithm and Scale angle adaptive mean shift algorithm through experiments. The results show that without increasing computational complexity, improved meanshift target tracking algorithm has higher accuracy, it can effectively eliminate the influence of background interference on target tracking, itcan adapted to the slowly changes of the background,at the same time can be adaptive to the scale shift and angle rotation of target,achieved good results.