Small Target Tracking Based on Histogram Processing  and a New Similarity Measure

Small scale target tracking is one of the primary difficulties in visual tracking. In this paper， we extended the framework of mean shiftbased small target tracking. First， we analyzed two major problems in small target tracking， namely tracking interrupt and target losing. Then the tracking algorithm was modified to address these problems. The main contributions of this paper contain: the histogram bins were reindexed to represent the target bins compactly; the target candidate histogram was smoothed if it is mismatch with the target model; a new similarity measure was proposed to improve the tracking accuracy and robustness， and the corresponding weight computation was derived. Several tracking experiments for real world video sequences show that the proposed algorithm can track the target effectively and accurately.