Video target tracking based on mean shift algorithm with Kalman filter

The mean shift based object tracking algorithm has been successfully applied in visual tracking due to its real-timeliness and robustness. The key module is mean shift iterations and it eventually converges to the object position in the current frame. In this paper, we prove its efficiency by experiments and discuss its weakness. An improved algorithm combining Kalman filter with mean shift is presented, aiming at basic mean shift algorithm fails to track fast-moving object.