Object Tracking Algorithm based on Improved Context Model in Combination with Detection Mechanism for Suspected Objects

Built upon the methodology of “spatio-temporal context”, a simple yet robust object tracking method is proposed for solving the occlusion problems in this paper. This algorithm makes full use of the context information of the object and its local background to calculate the features, which maximumlly improve the occlusion predictive response and recapture accuracy. Firstly, an early warning mechanism is adopted to realize the occlusion detection. Once the object is fully occluded, the object position with accurate motion information saved in the early warning is predicted and memory tracking model is used to delete the suspected object region, which reduces the matching complexity. Finally, a confidence strategy for similarity measurement is adopted to capture the suspected object when the object appears, and the optimal confidence is introduced to get an adaptive update model. Many simulation experiments in benchmark videos show that our proposed algorithm achieves more favorable performance than these existing state-of-the-art algorithms.