A method of athlete foul action recognition based on DTW algorithm
Weihai Zhong, Yanpeng Zhao, Cheng Yang, Chuan Wang · International Journal of Biometrics · 2025
There may be various foul actions in different sports events, which vary in form, speed, and rhythm, increasing the difficulty of action recognition. Therefore, a DTW algorithm based athlete foul action recognition method is proposed. Select the OV7670 camera to collect athlete action images, extract image contour information, and use it as a feature of athlete foul actions. Based on the collected images, evenly sample the contours of the images to obtain contour feature points. Optimise the DTW algorithm from three aspects: reference template modelling, distortion calculation, and determining the optimal path. Use the extracted action feature sequence as the input of the DTW algorithm to output the recognition results of athlete foul actions. The experimental results show that the distortion of this method is as low as 0.58, indicating that it can accurately identify athlete foul actions and has high application value.