Target salient confidence for visual tracking

In this paper, a novel visual tracking algorithm that uses the target salient confidence (TSC) is proposed. Two contributions are summarized as follows. First, we put forward a novel target salient confidence (TSC) model which combines the static saliency map (SSM) based on the selective visual attention model, motion attention map (MAP) and the target prior confidence (TPC). Second, we propose to use the target salient confidence for particle filter. It manipulates the distribution expressed by the particle cloud towards a better match with the target salient confidence model. In this way particle sampling can be locked on those regions with higher target salient confidence. Experiments in some video sequences show that the target salient confidence is useful for visual tracking and our algorithm is effective.