Research On Moving Object Detection And Tracking Based On Video Images

With the development of computer vision, pattern recognition and artificialintelligence, intelligent visual surveillance system has developed rapidly. Moving objectdetection and tracking based on video images have taken an important part in theintelligent video surveillance system, which are also the core technique. They are at thebottom of the whole visual surveillance system and are the basis of the subsequentprocessing.Based on the relevant research achievement, further study of moving objectdetection and tracking is done in this paper. Firstly, a brief review of the imagepreprocessing technology is given, focus on the image enhancement technique, using amulti-scale contrast enhancement technology based on Laplacian pyramid algorithm toimprove the algorithm in case of poor light processing. In the moving object detection,several typical detection methods are discussed. Their advantages and disadvantages areanalyzed. A moving object detection method of combing inter-frame difference methodand background subtraction method with block denoising method is proposed. Finally,the trajectory fitting prediction algorithm based on the first two moments is used totrack moving object.The video images collected from camera are used in the paper. In the matlabenvironment the tracking algorithm is simulated, and the experimental results show thatthe proposed method are effective and feasible.