Robust complete occlusion handling in adaptive template matching target tracking

摘要

Occlusion is a common problem in visual tracking applications. In adaptive template matching target tracking, characteristics of the occluding object may be erroneously incorporated in the new updated template. Such a corrupted template affects the tracker robustness and may lead to target loss. A novel technique is proposed for complete occlusion handling, in the context of a SWAD-based adaptive template matching tracker. A matrix of pixel weights is adopted, so separate parts of the template are updated at different rates and the overall structure of the target template is significantly preserved. Experimental results show that such a technique improves the robustness of the tracking algorithm in occlusion scenarios, for a successful recovery once the target becomes visible again.