Efficient multi-cue predictive tracking of multiple similar interacting objects

This work presents effective use of different object cue’s to handle interaction problem of multiple objects. We used traditional color histogram based particle filter to track each independent object. In case of object’s predicting region interaction, a sub-region selection method is used to select a set of probable sub-region. Finally, segmented motion with ellipse mask to each sub-region and object’s linear predictive location cue tracked interacting object correctly. Here, an approach is used that will give higher importance on better cue between those cue. We have experimented our method to similar color interacting random moving object and show better performance of our proposed method.