Robust Color Appearance-Guided Particle Filter for Object Tracking

This paper presented a color image tracking method. In order to implement an effective and robust tracking task, a novel approach of weighted color distribution appearance model based on color histogram is discussed, which takes into account the target´s shape and position of pixels as necessary factors in target model. Experiment result shows, in contrast with Mean Shift and conventional particle filter, the proposed method can locate object more accurately and have lower complication of calculation. Especially for some complex tracking conditions such as occultation illumination changes, cluttered backgrounds and the proposed method have a better robust.