Biomechanical analysis of shooting performance for basketball players based on Computer Vision

Abstract

Abstract Shooting technique is exceedingly crucial in basketball. In order to improve the shooting average, this study captured shooting technology action by computer vision technology, analyzed human posture in different gender and gave effective suggestions in routine training. Kinect Azure was used to collect the human skeleton information when the 10 players (5 males and 5 females) who experienced amateur basketball training shot. After smoothing and reducing noises of three-dimensional skeleton information, the kinematic parameters such as human joint angle and torso tilt angle were calculated by using Euclidean dot product formula and anti-triangular formula. The result showed that shooting average of male was higher than that of female (P<0.05). When the players squatted to the maximum extent, the angles of the knee and hip of female were significantly bigger than that of male (P<0.05). The torso was tilted to a certain extent, and the extent of female was significantly larger than that of male(P<0.05). In conclusion, by computer vision obtaining data collection, this study suggested that the players should strengthen the muscle and master the range of flexion degree which was suitable for their own strength. At the same time, so as to improve the shooting average, they should pay attention to the coordination of upper and lower limbs during shooting.