Recognition method of kicking and stride movements of Taekwondo athletes based on Gaussian mixture model

Abstract

In order to improve the accuracy of Taekwondo Athletes' motion recognition, a human motion recognition method based on Gaussian mixture model is proposed. This paper decomposes the structure of kicking and stride movement of human Taekwondo athletes to form a group of recognition targets, and investigates the changes of muscle surface under kicking and stride movement of Taekwondo athletes. The Gaussian mixture model is used to collect the shorttime feature images of kicking and stride movement of Taekwondo athletes, and the wavelet time-domain features of the images are obtained as the recognition vector space, The experimental results show that the method can effectively identify the movement characteristics and muscle changes of Taekwondo athletes, fully meet the research requirements.