5-5 Robust Tracking Method of Occluded Moving Objects Based on Adaptive Fusion of Multiple Observations

Abstract

This paper reports an occlusion-resistive tracking method of moving objects based on Kalman filtering. The filters adaptively fuse heterogeneous tracking strategies depending on the reliability of each strategy, and on the occlusion status. Experimental results with a soccer scene validates its robustness and feasibility for an object-based data broadcasting service