Hierarchical model based human motion tracking

Image sequence based tracking is the pivotal technique of human motion. In the paper, we first propose an appropriate human model and color model. Second, two approaches are proposed aiming at two different levels of color-block information including the boundary and the inner-area: the extraction of the boundary algorithm is based on the Robert operator and the clustering algorithm is based on self-adaptation. Third, we unite two regions, which are processed by different approaches. This step counteracts the ambiguity of obtaining information from each level. Finally, after the rough area of the color-block is obtained, the boundary of the color-block is determined by calculating the histogram of the X,Y coordinate of every point on the boundary of the block. Experimental results are presented