To Track The Moving Target In Intelligent Monitoring Technology Research

With the rapid development of economy and sociaty, people’s demand for informationtechnology and intelligent is increasing. The traditional manual monitoring methods are toodifficult to meet the monitoring requirements. So intelligent monitoring technology come intobeing. Without management, the intelligent monitoring system uses computer vision techniquesto automatically analyze video surveillance sequences and automaticiy identify specific targetsand abnormal behaviors in order to provide warns to managers. Intelligent monitoringtechnology which is paid extensive attention by reserchers at home and abroad, is applied in thefields of public safety, military, transportation and others. Target tracking is the core technologyof intelligent monitoring to guarantee the realization of the intelligent monitoring system, and ithas important research significance.At present, a lot of research work about moving target tracking technology has done by theacademic community, and a number of useful tracking algorithms were put forward. Accordingto the different characteristics of targets and the different principles of the methods, the algorithmof target tracking can be broadly divided into the tracking method based on motion analysis, thetracking method based on statistical model, the tracking method based on shape matching, thetracking method based on contour, and the tracking method based on region. The trackingmethod based on region which is relatively simple and has a small amount of computation, hasbeen widely applied in the field of target tracking. Mean Shift algorithm is one of the trackingmethod based on region. It has been widely used in target tracking field due to the characteristicsof real-time, robustness and easiness to implement. Hower, the tracking algorithm based on theMean Shift has bad ability to adapt fast motion of the target and the problem of serious occlusion.If such problems arise, there will be a big deviation with the actual position of the target in theresult of tracking, even appears tracking failure. Based on the study of such problems, a series ofwork have been done as follows.Firstly, methods of target detecting and tracking are researched in this paper. The basic ideasand general process of detection algorithms and tracking algorithms are analyzed, and theworking principle of the moving target tracking methods based on Mean Shift and otheralgorithm are mainly explored. Secondly, the defect of the Mean Shift algorithm are analyzed.For the Mean Shift tracking algorithm do not adapt to problems of fast motion of the target andserious occlusion, a target tracking method based on the Mean Shift and trajectory prediction isespecially advanced. A measure to deal with occlusion is researched, and related experimentshave been done carefully. In order to analyze the advantages of new algorithm, the results ofexperiments are compared with that of the traditional Mean Shift algorithm. Thirdly, according to the theory of fitting curve based on least squares method, the fitting curve trajectory prediction(FCTP) model is proposed. Then, FCTP model is combined with Mean Shift algorithm that hasadaptive tracking window to get a new algorithm of moving target tracking. According toexperiments, it is known that the proposed algorithm not only can effectively deal with the scalechange of target, but also can overcome the problem of fast moving of target and seriousocclusion in the course of moving, thus ensuring the tracking accuracy of the algorithm.