The Algorithm Research Of Tracking Based On Switching Particle Filters

Moving object tracking in video assembles advanced technologies in many fields such as image processing,pattern recognition,artificial intelligence,automatic control computer application,physiology,physics and mathematics, et al. It has many excellent advantages of visual scene, anti-interference and cost-effective, and has been widely applied to military surveillance, traffic control, machine intelligence, medical diagnosis, and so on. With the continuous demand of complex environment applications, besides introducing new technology, how to improve accuracy and robust of the existing tracking algorithm is the focus of the current work.The research in this thesis focuses on two aspects of the video tracking. One is the feature description of the object. The other is the framework based visual tracking, particle filter.Firstly, In order to deal with the robust feature description of object, this paper proposes a new histogram-HOGC.HOGC combines the color histogram and grade histogram. The two histograms can complement to each other. The color histogram is unaware to rotation while the grade one is unaware to color. So HOGC is a robust feature of object in visual tracking.Secondly, this thesis proposes a new particle filtering scheme, called a switching particle filter, which allows robust and accurate visual tracking under typical circumstances of real-time visual tracking. This scheme switches two complementary sampling algorithms, Condensation and Auxiliary Particle Filter, in an on-line fashion based on the variance of the current estimated target state. The accuracy and robustness of the switching scheme were evaluated using real visual tracking experiments.