Robust human body tracking using sparse representation

Robust human body tracking is a challenging problem due to the presence of noise, occlusion and variation of illumination under a complex environment. A variety of tracking algorithms have been proposed to overcome these difficulties. In this paper, we propose a robust human body tracking method by casting tracking as a sparse approximation problem in a particle filter framework, namely robust sparse representation tracker (RSRPF). The problem of occlusion, corruption and illumination changes can be well addressed through a set of trivial templates. To further improve the tracking performance, non-negative constraints are enforced to filter out clutters which negatively resemble tracking targets. We test the proposed approach on three challenging sequences involving heavy occlusions, drastic illumination changes, and background clutter. Experimental results demonstrate that our tracking method is robust in handling occlusion, background clutter and illumination.