Research On Tracking Algorithm For A Target On The Ground

This paper does some research work on the tracking algorithms toward targets on the ground, which involved all the key stage of correlation tracker. Through analyzing the model of correlation tracker, this paper put forward some improving algorithm toward some problems that restrict the application of correlation, such as the drift of the template and the real-time performance, etc. At the same time, This paper put genetic algorithm into use of correlation tracker, apply which to the problems difficult to work with regular algorithm, such as the recognition of objects, the automatic object capture, and the matching of rotate image. Finally, we perform the trial of picking up the feature argument with genetic algorithm. This paper consists of two parts, the first part is a new method of correlation tracking, including the description of correlation tracking algorithms, template pick-up , the fast matching method and the following processing of correlation tracking. At the same time, we compare the new algorithm with the classic algorithms and which proved that all the performance of ours are prior to the old one. The second part describes a method of using genetic algorithms to solve some problems that difficult to work with the regular algorithms, including the correlation match of rotated image, the automatic capture of the lost target and correlation feature matching algorithms.