基于单目视觉的网球运动员持拍手运动轨迹检测方法

Detection Method for Trajectory of Tennis Players’ Hand of Holding Racket Based on Monocular Vision

网球运动员持拍手动作变化速度较快,导致其运动轨迹检测效果较差,为此,提出基于单目视觉的网球运动员持拍手运动轨迹检测方法。利用帧差法分割运动视频图像,得到持拍手候选区域,通过分割肤色区域获取最终的持拍手图像,完成运动轨迹检测。实验表明,该方法可有效采集网球运动员运动视频图像,可精准检测持拍手运动轨迹,该方法运动轨迹检测的AUC值均较高,具备较高的检测精度。

The posture speed of tennis players’hand of holding racket is fast,which leads to the poor effect of their movement track detection.Therefore,a method for trajectory detection of tennis players’hand of holding racket based on monocular vision is proposed.The frame difference method is used to segment the motion video image to get the candidate area of their hand of holding racket,and the final image is obtained by segmentation of the skin color area to complete the trajectory detection.Experimental results show that this method can effectively collect the motion video images of tennis players and accurately detect the motion track of their hand of holding racket.The AUC values of the motion track detected by this method are all high,having high detection accuracy.