An Approach Based on Mean Shift and Background Difference for Moving Object Tracking

The traditional fixed bandwidth Mean-Shift Tracking algorithm can not have an effective tracking for any changes in targets. Based on the current methods, the tracking are particularly effective on gradually increase or decrease in size, there has no Mean-shift Tracking fit the size in any change. On the basic of this, we propose a combination of traditional algorithm bandwidth adaptive algorithm. Background difference method previously only used for target detection, and I will judge it as a kind of generation method of the determination conditions which can determine the traditional kernel-based target localization method and automatic bandwidth selection method combined. In this way, it makes the new algorithm can be adaptive bandwidth scale changes in targets. Any change to the pedestrian scale of the tracking experiments confirmed the effectiveness of this algorithm.