Image segmentation and proto-objects detection based visual tracking

Abstract Considering the phenomenon that the traditional visual tracking methods are prone to drift in the process of tracking, we propose a visual tracking approach based by image segmentation and proto-objects detection in this paper. Firstly, according to combining several segmentation approaches and topic model, the proto-objects are detected for a given image sequence. Then the final target position is given on the basis of proto-objects information. The coinstantaneous tracking of these proto-objects improves computational efficiency and tracking robustness. Detailed experimental results show that the proposed method perform better than the state of the art in anti-occlusion and illumination change.