Color and shape are the two main features of an image target.In order to improve tracking accuracy and robustness,a new multi-feature fusion tracking algorithm fusing color and shape features of the image target is put forward.The histogram model of space color probability based on HSV space,and the Hu invariant moment model after the gray-scale transformation are adopted respectively for real-time target tracking.And then an adaptive weighted method is adopted to confirm the final tracking location of the image target.Two experiments demonstrate that the proposed method has very strong robustness under the circumstances of complex background interference and to the change of color,scale and brightness.At the same time,the algorithm improves the tracking results and the effective rate of tracking.