摘要
传统的网球运动员发球技术动作图像分割方法,在发球动作的轮廓提取上不够精确,导致图像分割精度较低。因此,提出基于深度掩码的网球运动员发球技术动作图像分割方法。首先,输入需要配对的网球运动员发球技术图像,生成伪标签,将具有形状先验信息的标签作为辅助掩码,对图像进行标注处理,使用深度掩码,根据技术动作图像的标注结果,对具有完整轮廓的网球运动员发球动作类目标进行阈值分割,精确提取发球动作的轮廓,并使用Canny边缘支持方法,构建轮廓的边缘矩阵,形成分割后的网球运动员发球技术动作图像。进行实验,结果表明,本文设计的方法对网球运动员发球技术动作图像分割精度达到97.61%,能实现较高精度的发球技术动作图像分割,应用效果较好。
The traditional image segmentation method for tennis players'serve techniques is not accurate enough in extracting the contour of serve movements,resulting in low image segmentation accuracy.Therefore,a depth mask based image segmentation method for tennis players'serve techniques is proposed.Firstly,input the image of the tennis player's serving technique that needs to be paired,generate a pseudo label,use the label with shape prior information as an auxiliary mask,annotate the image,use a depth mask,and based on the annotation results of the technical action image,perform threshold segmentation on the tennis player's serving action target with complete contour,and extract the contour of the serving action.Use the Canny edge support method,and Construct an edge matrix of the contour to form a segmented image of the tennis player's serve technique movements.The experimental results show that the designed method in this paper achieves a segmentation accuracy of 97.61%for the image of tennis players'serve technique movements,and can achieve high accuracy in image segmentation of tennis players'serve technique movements.The application effect is good.
作者
柏宁
孙永梅
BAI Ning;SUN Yong-mei(Department of Physical Education,Anhui University of Finance and Economics,Bengbu 233000,China;College of Physical Education,Anhui Normal University,Wuhu 241002,China)
出处
《安徽师范大学学报（自然科学版）》
2024年第1期69-74,共6页
Journal of Anhui Normal University(Natural Science)
基金
安徽省哲学社科孵化项目(AHSKF2021D23)
安徽财经大学校级重点教学研究项目(acjyzd2022028)
安徽省高校自然科学研究重点项目(KJ2019A0497)
安徽省高校人文社科研究重点项目(SK2020A0041).
作者简介
柏宁(1986-),男,安徽蚌埠市人,硕士,副教授,研究方向为体育产业与教育.
