A robust object tracking method combining shape descriptor and adaptive background camshift

Feature-based object tracking is one of important object tracking method. Features fusion can improve can improve the tracking performance. This paper proposed an algorithm which combines Hu moments and color features. Moreover, this paper integrates the ABCshift algorithm to overcome color features drawbacks which easily lead to loss of target object when background exists clutter. To test the effectiveness of proposed algorithm, two challenging video sequences have been tested on this algorithm and other trackers. Experimental results demonstrate that the algorithm has strong robustness and improves the tracking performance.