Object tracking by supported particles
Abdullah Nazib, Chi-Min Oh, Chil-Woo Lee · 2014
Object tracking is still remains as a challenge to the computer vision community. Several methods have been proposed until now to track. One renowned method to track is particle filter, a probabilistic model that predicts object position based on recursive Bays formula. In this paper, we present a particle filter based object tracking method, where a set of contextual points is used to support particle filter. The primary contribution of this proposed idea is to the use of context information as the support for particles and also to use those context points as observer to observe angular velocity of the tracked object with respect to each context points. In this paper we call those points as supporter. The angular velocity with respect to support points are then used to handle occlusion.