The Research On Techniques Of Tracking Video Moving Target

With development of computer technology in recent decades, tracking target of video system has been widely applied in various fields. It is difficult to track target with the complex background and the change of the target size. There are many problems which can't be resolved, although people have been in-depth researches and proposed a variety of effective methods.First of all, the thesis summarizes the research background and significance of the topic, then analyses the development trend and current situation. Finally, it gives the brief illustrations about the emphasis of research and every step. The emphasis of research is how to deal with the change of target size and resolve the occlusion problem because it can obviously influence the target tracking.The moving target detection is the base of target tracking in video. It contains image preprocessing, the methods of the target detection and morphological processing. Besides, the thesis also gives some experiments about the target detection. The experimental results show that the methods of target detection are effective. So it can give the exact model for the target tracking.In terms of the target tracking, first of all, the thesis analyses the theory of the target tracking which contains the traditional algorithms, the forecast algorithms and the similarity measure methods. Then it researches the MeanShift algorithm and Kalman filter. Finally, it proposes some new methods using Kalman filter in order to remedy the shortages of MeanShift algorithm. It contains the method of forecasting the target position, the method of updating the target model and the method of resolving the occlusion problem. The experimental results show that it can track the moving target with fixed size well when the target is occluded.To resolve the change of the target size, the thesis proposes a new method of forecasting target size which combines the target information measure and Kalman filter. The experimental results show that it can track the target well with the adaptive size when the target size has changed. Besides, the thesis proposes a new method to deal with the target occlusion which combines the target information measure and the anterior method of dealing with the target occlusion. The experimental results show that it can also select the proper target size well when the target is occluded.