Robust Compressive Tracking Under Occlusion

In this paper, we present a robust and fast object tracking algorithm based on sub-region classifiers and compressive tracking. Compared with the original CT algorithm, the tracker can improve the robustness to occlusion, especially long-term occlusion. Firstly, the target region is divided into four sub-regions in a fixed mode. Then a simple but feasible classification and update strategy is used for these sub-regions classifiers. On the assumption of rigidity, the final location of the target can be evaluated by these sub-regions classifiers. The experiments on many challenging image sequences demonstrate that the proposed method achieves more favorable performance than several state-of-the-art tracking algorithms in terms of speed, accuracy and robustness.