An advanced association of particle filtering and kernel based object tracking

The association approaches of particle filter (PF) and kernel based object tracking (KBOT) are widely used in visual tracking. Specially, a compact association approach is proposed, which is based on an incremental Bhattacharyya dissimilarity (IBD) and condition number. It is advanced approach, but this paper found that it cannot guarantee the stable tracking and the high accuracy of tracking in any cases. To solve these problems, we first introduces an asymmetric incremental Bhattacharyya similarity (AIBS) instead of IBD. AIBS is defined by incorporating an asymmetric incremental similarity matrix (AISM) and enables to ensure the stability of tracking. Then, we propose a boosting–refining approach, which is boosting the particles positioned at the ill-posed condition instead of eliminating the ill-posed particles to refine the particles. It enables the estimation of the object stare to obtatin high accuracy. Also, We propose also the resampling-refining algorithm to advance the performance of the framework based on the association of PF and KBOT. Finally, we test the stability and the accuracy of the association approaches that we proposed in this paper, on a synthesized image sequence and several real image sequences. Experimental results demonstrate that our approaches have the promising discriminative capability in comparison with other ones.