Edge detection techniques assisted target tracking algorithm

Abstract

This paper proposes an improved approach for target tracking. The new approach addresses the tracking failure issue of Mean-shift algorithm when the dimension of an object changes over time. Object edge detection is implemented into the tracking process. The target can be located more accurately with an adaptive correction system based on the information of object edges. In addition, an improved edge detection algorithm is also studied to overcome the problem of low tracking accuracy in traditional methods. The experiments demonstrate that, compared with the traditional Mean-shift algorithm, the proposed approach can significantly improve the performance in tracking accuracy.