Multiple Moving Targets Tracking Research in Cluttered Scenes

Traditional particle filter (PF) will have bad effect when the target's color is similar to the background or the target is blocked, and an improved tracking algorithm is presented in the framework of PF and DSmT. According to the location cue, a new tracking approach is given to establish the model of multiple moving targets based on information fusion. Experimental results demonstrate that the proposed approach can accurately track the intersecting targets in cluttered scenes, and high conflicting information is dealt with effectively. So, the introduced method improves the reliability and rationality of multiple moving targets tracking, and a robust tracking result is obtained although evidences conflict another one highly.