Improved Anti-Occlusion Target Tracking Algorithm Based on Compressive Particle Filtering

Abstract

In order to resolves the problem of occlusion in process of target tracking, an improved anti-occlusion tracking algorithm was proposed in this paper based on compressive particle filtering (CPF). Compressive sensing theory was introduced into particle filter (PF) framework to ensure the instantaneity of tracking. We apply the histogram with spatial information and sub-part matching ideas in the compressive particle filtering algorithm to enhance the robustness of tracking when the target was blocked by barriers. In this approach, we adopt different strategies to tracking target when the target was occluded or not. When the target was occluded, tracking it by compressive particle filtering algorithm based on sub-part matching and updating the target templates to fits the change of target appearance, otherwise, tracking it by the general compressive particle filtering algorithm. This approach bring about better robustness and tracking speed compared with the particle filtering algorithm and compressive tracking algorithm.