Adaptive target color model updating for visual tracking using particle filter

Color histogram distribution is robust against non-rigidity, scale and rotation. Color-based particle filtering is one of the most successful object tracking paradigms. But visual tracking in real world conditions such as changing illumination and poses is still a challenging job. In this paper, we develop a color histogram based particle filter tracker with adaptive target model updating. The proposed approach adds two auxiliary variables in the particle state space. These two auxiliary variables control the updating speed of the color observation mode, and are also estimated in the sequential Monte Carlo framework. This algorithm has been tested on real image sequences and accurate tracking result has been achieved.