An adaptive particle filter tracking algorithm based on multi-information fusion

Abstract

Aiming at the problems caused by target shape and scene environment changes in video target tracking, a novel adaptive particle filter tracking algorithm based on multi-information fusion is proposed. To improve algorithm adaptability to environment changes, MPC distance, color histogram and edge orientation histogram are comprehensively considered during the particle weight updating process. An adaptive strategy for particle filter parameters and template updating is presented to reduce the influence of target shape changes on the tracking results. Experimental results show that the proposed algorithm is of effectiveness.