Visual Tracking Using the Kernel Based Particle Filter and Color Distribution

In this paper we propose a new approach for tracking an object in a video sequence. Our tracker is mainly composed of object modeling and particle filtering based on kernel methods. First, to overcome the problem of appearance changes, we model the target by computing kernel density estimation of color distribution of interesting objects. To improve the performance of tracker based on the classical particle filter, we employ a kernel based particle filter that uses a broader kernel to form visual tracker. Experimental results show that the proposed method can obtain the superior performance to the tracker using the generic particle filter.