Human Motion Tracking via the Local Dissimilarity Map

In this paper, we propose a robust visual tracking algorithm based on local dissimilarity map (LDM) and Kalman filter (KF). Firstly, we model the motion model component of the proposed tracker by using the KF. Then, we apply the LDM into the object matching model to measure the local dissimilarities between the target and the sampled candidates in each frame of the given image sequence. Experimental results on several image sequences illustrate that the proposed method performs well in several challenging aspects of real world scenes.