Particle Filter Based Target Tracking In Video Processing

The technology of moving objects tracking is a meaningful subject in computer vision. It has a lot of applications in many fields such as missile controlling-and guiding, video surveillance, medical treatment, traffic control, etc. Since IR(Infra-red) imaging possesses many advantages, it has become the trend and direction for current development and study of tracking technologies worldwide.Kalman Filter and Extended Kalman Filter are the most typical filter algorithms in target tracking domain. Although these two methods own pretty good filtering performance when system noise and observation noise are Gaussian, their filtering performance will descend or even diverge when non-Gaussian distribution occurs. Particle filter realizes recursive Bayesian filter via Monte Carlo simulation. The method is suitable for any non-linear system that could be represented with state model. It is more practical than conventional Kalman filter and its precision could approach optimal estimation. In this paper, research on particle filter algorithm for IR image target tracking is carried on under the main background of nonlinearity, non-Gaussian noise. Through analysis of the basic theories of Kalman filter, correlation tracking and particle filter algorithm and study on remaining problems of these algorithms, we discuss an approximate optimal importance density function particle filter and evolutional particle filter, analyze the disadvantage of the complicated calculation and object drifting out of the reference template, and eventually propose a target tracking algorithm based on particle filter and correlation tracking. The simulation results shows that the new tracking algorithm can primely solve the IR image target tracking and has less quantity of computation and high robustness.