Particle filter tracking method based on dynamic template update strategy

Abstract

Traditional tracking algorithm is not compatible between robustness and efficiency, and at present, the template update strategy is not robust to target appearance changes. Therefore, the paper presents a dynamic template-update method as well as embeds it into the particle filter tracking method. By incorporating the original information into the updated template, the presented algorithm has the natural ability of anti-drift. Besides, according to the variety of each component in template, a dynamic update strategy is proposed to adjust the updating weights adaptively, thus boost the performance of efficiency under fast or nonlinear appearance variation. Experimental results show the feasibility of the proposed algorithm in this paper.