Adaptive Fragment Multi-Target Tracking in Occlusion Scene

Multi-target tracking in occlusion scene is a challenging task. This paper proposes a multi-target tracking method based on adaptive fragment to complete it. The method adaptively divides each target into a few of fragments on the basis of its multi-channel gray projection. The similarity between each fragment and its model is calculated by weighted Bhattacharyya coefficient. Fuzzy C-means method is used to cluster the particle sets of each target, and the optimal state estimation of the target is obtained from these sets. By updating fragment weight of target model, the method can dynamically deal with occlusion problem. Experiments show that the proposed method can improve the accuracy and robustness of multi-target tracking in occlusion scene.