Human tracking using multiple-camera-based head appearance modeling

We propose an appearance-based method for tracking motions of multiple persons using an asynchronous multiple-camera system. In the proposed method, the head appearance of each target person is dynamically modeled using multiple-camera-based observations. Observed color (texture) information and the related reliability values are stored in the head model as a set of color patches distributed in 3-D space. The head appearance model can be easily produced from any view direction by using the stored information. Head-appearance-based tracking is, especially, useful for reducing the impact of human-human occlusions. We applied the proposed method to a human interaction scene that contains small inter-person distance where the motion-based (silhouette) method fails due to occlusions. Experimental results show the effectiveness of the proposed method.