Mean-Shift tracking algorithm based on Kalman filter using adaptive window and sub-blocking

To reduce the tracking errors caused by high-speed motion and variable motion in the process of moving target tracking, a novel Mean-Shift tracking algorithm based on Kalman filter using adaptive window and sub-blocking is proposed in this paper. Moving target´s utmost position is predicted by combining Kalman filter and historical information, which is used as the initial position. During describing feature model of target and candidate regions, they are blocked and each sub-block region is processed by reducing RGB interval, by which computational efficiency will be improved. Finally, window bandwidth will be enlarged and reduced according to Bhattacharyya coefficient, and it achieves accurate moving target tracking by adaptive window. The comparison experiments of the Coastguard standard image sequence and car image sequence demonstrate that the proposed tracking algorithm is insensitive to high-speed motion and variable motion of moving target, and it has better tracking performance.