Visual tracking via guided filter

In this paper, we propose a novel tracking algorithm based on an explicit image filter - guided filter. The guided filter utilizes the structure in the guidance image and performs as an edge-preserving smoothing operator. In our work, we treat the target as the guidance and the incoming candidates are filtered depending on the similarity between the guidance image and each input. The edge-preserving smoothing property depending on the guidance image is a critical advantage for object tracking. First, the guided filter can help to pick out the valuable candidates and make the inaccurate ones blurry so that the tracker can distinguish the target from numerous bad candidates easily. Besides, the filtering process can recover the content of the target being occluded according to the guidance image, which can help to alleviate the drifting problem effectively. Eventually, to generate a robust tracker, we take advantage of the combination of positive and negative templates to conduct effective sparse representation. Experimental results show that our algorithm outperforms relative trackers.