Studies On Moving Objects Detection Technology In Intelligent Video Surveillance

Video surveillance system is a crucial issue in the field of computer vision,pattern recognition and artificial intelligence, which is the important high-tech measures and technology to guarantee the public security in modern society. One of the main task of intelligent video surveillance technology is to monitor whether there are new targets appear, the basic content of which can be divided into two areas, first to detect the target in the video sequence, followed by the target extracting. The relevant theory and our new idea in intelligent video monitoring based on computer vision is described in this thesis, which consist of six chapters.Firstly the paper summarize the domestic and international status, the theories and technology of moving target s detection in intelligent video surveillance, as well as research several mainstreams of moving object detection and segmentation methods.Then four background modeling algorithms are widely researched and analyzed based on extensive experiments. The comparison conclusions are presented. An improved background modeling algorithm is proposed based on above experiment and research, which combined the brightness and energy information of the pixel in frame difference image, not only the complexity of the algorithm is less than Gaussian mixture background modeling methods, the advantage of Gaussian mixture model background modeling extracting a clear foreground is kept. The relevant experiments, conclusions and analysis are given in end of the chapter.Thirdly the paper describes a new method of moving target detection based on the distribution of frame difference information variance, using variance function the pixels belonging to moving targets could be determined; then compose the pixels connected to each other to the different target regions, after the sub-region within the target being filled ,the complete target area could be abstained; Finally the segmentation moving target is detected out, and the experimental results and conclusions are given and analyzed. Finally, the overall of this thesis, conclusions and future work in intelligent video monitoring are given.