Remarks on Simple Motion Capture Using Heuristic Rules and Monte Carlo Filter

This paper investigates a simple human motion capture method that estimates the human body´s representative points by using analysis of the human silhouette extracted from the camera image. The estimation method utilizes both the heuristic extraction rules of the representative points using a priori knowledge of human body and the contour analysis of the human silhouette. The 2D coordinates of the representative points, such as head, neck, hands, toes, joints of elbow, and joints of knee, are located by using the estimation method. To overcome self-occlusion problems of the hands, colour template matching of the hand is utilized and the feature tracking based on Monte Carlo filter is introduced into the estimation method. Experimental results show both the feasibility and the characteristics of the proposed method to achieve simple human motion capture in real time.