Abstract: A part-based tracking method is proposed to address the occlusion issue in object tracking by using parts of the object as the tracking objects.Parameters of each part are predicted by an independent Kalman filter and corrected by kernel-based histogram which is used to describe the part.Experimental results show that this method can effectively overcome problems caused by the occlusion，the non-rigid deformation and the relative motion in a tracking object.Meanwhile，they also prove that the method achieves good performance in real-time object tracking.