Research Of Target Tracking Based On Classifying Multiple Model Algorithms

The original purpose of target tracking was to solve problems in military, and it was developed rapidly. More and more researchers joined this field and made great contributions. In military, target tracking was widely used in Aerial reconnaissance and warning, Ballistic missile defense, Battlefield surveillance etc. Target tracking also was used in our live, Air traffic control, intelligent vehicle system, Traffic navigation, Robot vision etc. The target tracking technology was not only tracking the target in linear condition. As the industrial technology developed, it had to track target in more and more difficulty environment, and most kind of moving target.This thesis analyzed multi-model methods of target tracking, and found their applicable status and environments. According to their shortcomings, we found a new method CMM which can improve the filter precisions while target moving stability. A lot of works had been done in the thesis:(1) Analyzed many target tracking methods, and classify them into three groups, which is weak maneuvering target filter, medium maneuvering target filter and strong maneuvering target filter. They are the main filters in CMM.(2) Did a lot research on IMM, and found the weakness that when the target moving stability the filter precision will reduce. In order to solve the problem, a new method CMM was designed in the thesis which will use only one filters for each target when the object moving stability.(3) Researched a lot of filter methods which based on Kalman filter and find a way to detection the maneuvering of the moving targets.(4) The thesis researched the realization of CMM, and a lot of experiments were done to check the maneuvering detection and classification. Simulate target tracking and the result was consistent with the expected.The research shows that CMM could get a good filter precision while the target moving stability. When it was used in maneuver target tracking, the maneuvering of the target would be detected accurately and the classification was absolutely correct.