Abstract:

Covariance tracking has achieved impressive successes in recent years due to its competent region covariancebased feature descriptor. However, its bruteforce search strategy is still inefficient.A generalized,adaptive covariance tracking algorithm is proposed,which uses novel integral region computation and occlusion detection. The integral region is much faster and adaptive to the tracking target and tracking condition.The adaptive search window is adjusted by simple occlusion detection. The integral image and the global covariance tracking can be seen as a special case of integral region and the proposed algorithm,respectively.The proposed algorithm unifies the local and global search strategies in an elegant way and smoothly switches them according to the tracking condition judged by occlusion detector.It gets much better efficiency and robustness for distraction and stable trajectory by local search in normal steady state, and obtains more abilities for occlusion and reidentification by enlarged search window (until to global search) in abnormal situation at the same time. Experiments on many video sequences show that the proposed algorithm has excellent target representation ability,faster speed,and more robustness. 

Key words: covariance tracking;integral image;occlusion detection;integral region;search window