Complex Form of Local Orientation Plane for Visual Object Tracking

Object tracking suffers from appearance change of the target caused by heavy occlusion. In this paper, we propose a complex form of local orientation plane (Comp-LOP) for visual object tracking. CompLOP is a simple but an effective descriptor for object tracking, which is robust to occlusion. It effectively considers the spatiotemporal relationship between the target and its surrounding region in a correlation filter framework by the complex form, which successfully deals with the heavy occlusion problem. Moreover, we perform scale estimation to treat scale variation of the target for better tracking performance. Besides, we automatically determine an appropriate size of the search region based on the entropy ratio of the target to the background. Experimental results demonstrate that the proposed method achieves good object tracking performance in accuracy, robustness, and computational efficiency.