Object Tracking Algorithm Based on CamShift with Dual ROI and Velocity Information Fusion

In order to deal with the tracking divergence and the recapturing failure after occlusion of the continuously adaptive mean Shift algorithm(CamShift) to track objects passing the background with similar colours to them,an improved CamShift algorithm with dual region of interest(ROI) and velocity information fusion is proposed to track moving objects.The main idea of the algorithm is to divide the single ROI which is used to specify the region to be tracked in CamShift into two sub ROIs,of which one is the primary tracking region and the other one is the auxiliary tacking region.For each of these two sub ROIs,a CamShift tracker is designed respectively.Through the coordination of these two CamShift trackers in the process of tracking,the tracking robustness of the algorithm is enhanced and the interference problem due to similar colour in the CamShift is solved.Meanwhile,the velocity information of the tracked object is introduced into the CamShift tracker to enable the algorithm to relocate the tracking target after occlusion.Experimental results demonstrate that the proposed CamShift with dual ROI can effectively track the object even when it is affected or occluded by things with similar colours.