Study on target tracking algorithm integrating color and contour features

Abstract

Aiming at the instability of traditional color feature tracking algorithm in some complex scenes, an object tracking algorithm with integrating color and contour features is proposed. The algorithm still uses the traditional particle filter to predict the target, but in iteration every time, when searching the best position of the target, the color and contour information of the target can be adaptively combined. Experiments show that it shows good effectiveness and robustness compared with the traditional algorithms when the target is interfered between similar colors or the target is partially obscured.