Research On Moving-Objects Tracking In Intellectualized Video Surveillance System Based On Network

Video surveillance is playing an increasingly important role in the national economy. As the rapid development of computer, network and image processing technologies, the video surveillance has made a great progress. Intelligentizing is one of directions which the modern video surveillance system towards. The critical of Intelligentizing is how to deal with the image series from video, detect the moving objects and track them. This thesis is mainly about that. This paper describes the normal procedure of objects tracking in video surveillance system. When the camera is fixed, there are two basic methods to track objects: background subtraction and two consecutive frames subtraction. As to the background subtraction, we offer a few ways to build the background model and analyze the method to treat with the sudden change of light. Due to the moving objects may have shadows, the author puts forward a way to clear the shadows using a grey threshold. Article puts forward one combination of background subtraction and two consecutive frames subtraction to track objects and have given the experimental result. As to the more complicated situation of moving camera, the article analyzes some tracking technologies too. There are some good methods, such as set up movement model of the camera and piece-matching. The thesis analyses and compares many kinds of methods of setting-up camera moving-model. After analyzing a method---background moving compensation, the paper puts forward one way based on matching to track moving targets when camera is moving, and provides the experimental result. On the basis of the consideration of the real-time in monitoring system, this text has also analyzed a kind of method of tracking based on the flow of video. This kind of method utilizes movement vector information included in the flow and it can realize faster tracking than others. In a word, this thesis makes a deep analysis and research into moving objects tracking towards various kinds of actual conditions in surveillance system. At the same time, the author has put forward his own methods, carries on experiments on a lot of processing course and gets kind experimental result.