Abstract
The existing recognition methods of complex human movements in Wushu have the problem of imperfect kinetic energy model, which leads to low recognition accuracy. A complex human motion recognition method based on bone point features is designed. Identify martial arts movement posture, combine the upward movement of human center of gravity trajectory, establish the kinetic energy model of joints according to the positioning results of extremity points, set the threshold of local spatial differences of human bones with the central node of hip joint as the center point, avoid overcalculation, and optimize the complex motion identification process by combining the characteristics of bone points. Experimental Results. The correct rate of different types of actions identified by this method is 90.1% and 92.7%, and the identification time is 1.2 s and 1.41 s, which shows that this method can identify actions quickly and effectively by combining the feature information of bone points.