State-space recursive least-squares based object tracking

Real-time tracking of non-rigid objects is a challenging task. The main objective is to accurately localize the target object in each frame. This in turn necessitates modeling of trajectory of the object. In this paper, we propose to use state-space recursive least-squares (SSRLS) for this purpose. Kernel based mean shift tracker has been used for the actual localization of the target in a frame. The combination of the two algorithms results is an improved performance and reduced search time. The proposed tracking scheme is robust in camera motion, partial occlusions, clutter, and target scale variations.