Research On A New Robust Visual Object Tracking Algorithm Based On Correlation Filter

For efficient and robust,the correlation filter algorithm has been widely used in object detection,object tracking and so on in recent years.The correlation filter based tracking algorithm,which can run in real-time,is robust in rotate,scale changes and partial occlusion.In the long tracking process,a single appearance model is difficult to represent multiple different states of the target,which will lead to tracking failure.To solve the problem of the distractors in tracking,an algorithm of a correlation filter based tracking algorithm with distractor detection(DDT)is proposed.Taking advantage of the constant of the relative position and the target appearance between the frames,a discriminative model is proposed to optimize and improve the target location strategy.Experimental results show that the use of the distractors' information has improved the trackers' robust in complex scene.To presentation the different states of the target in the long-term tracking task,a multi-tracker modal tracker(MTMT)is proposed.In the process of tracking,use the PSR as the basis for judging,dynamic create and choose the right tracker modal.Finally,experimental results show that MTMT improves the tracking performance in long time occlusion and large deformation scene.