摘要
研究分析763段高水平羽毛球比赛视频,HRNet-W48网络构建姿态估计系统PCK指标达93.7%。测量结果显示击球前肩关节外展角度139.6°±8.3°,肘关节伸展角度143.8°±6.7°,腕关节角速度1257.4°/s±123.6°/s。K-means聚类识别三种技术:肩主导型42.3%,前臂主导型35.7%,综合协调型22.0%。随机森林算法评分模型准确率89.3%,腕关节角速度贡献25.4%。世界排名1-10位选手腕关节角速度1342.5±98.7°/s显著优于11-20位选手1172.3±111.5°/s(p<0.05)。亚洲选手肩关节活动范围大,欧洲选手肘关节伸展速度高12.3%。综合协调型击球速度308.1±12.5km/h领先肩主导型8.7km/h。中国队60.7%选手采用前臂主导型,印尼队57.5%使用肩主导型,丹麦队41.3%属综合协调型。世界排名前五选手91.4%属综合协调型。顶级选手四年间技术向综合协调型转变增14.6%。躯干旋转角度与击球精确度呈正相关(r=0.67,p<0.01)。高水平选手重复执行角度波动低28.4%。肩-肘-腕协同指数高排名选手0.87±0.06优于低排名选手0.74±0.09(p<0.01)。研究建议强化三关节协同性,根据身体特点选择技术模式,注重击球点高度控制,建立个性化技术数据库。
This study analyzed763 high-level badminton match videos.Apose estimation system was constructed usingthe HRNet-W48 network,with the PCK index reaching 93.7%.The measurement results showedthat before hitting the shuttlecock,the abduction angle of the shoulder joint was 139.6°±8.3°,the extension angle of the elbowjoint was 143.8°±6.7°,and the angular velocity ofthe wrist joint was 1257.4°/s±123.6°/s.Threetechniques were identified throughK-means clustering:shoulder-dominated type(42.3%),forearm-dominated type(35.7%),and comprehensive coordinatedtype(22.0%).The accuracy of the scoring modelusing the random forest algorithm was 89.3%,and the angular velocity of thewrist joint contributed 25.4%.The angular velocity of the wristjoint of players ranked 1-10 in theworld was 1342.5±98.7°/s,significantly better than thatof players ranked 11-20(1172.3±111.5°/s,p<0.05).Asian players had a larger rangeof shoulder joint movement,whileEuropean players had a 12.3%higher elbow joint extension speed.Thehitting speed of the comprehen sive coordinatedtype was 308.1±12.5 km/h,8.7km/h faster than that of the shoulder-dominated type.60.7%of Chinese players adopted the forearm-dominated type,57.5%of Indonesian players used the shoulder-dominated type,and 41.3%of Danish players belonged to the comprehensive coordinated type.91.4%of the top five players in the world belonged to the comprehensive coordinated type.Over the past four years,the technique of top players has shifted towards the comprehensive coordinated type,with an increase of 14.6%.There was a positive correlation between the trunk rotation angle and hitting accuracy(r=0.67,p<0.01).High-level players had a 28.4%lower fluctuation in repeated execution angles.Theshoulder-elbow-wrist coordinationindex of high-ranked players was 0.87±0.06,better than that of low-ranked players(0.74±0.09,p<0.01).The research recommends strengtheningthe coordination of the three joints,selecting technical patterns according to physical characteristics,paying attention to the control of the hitting point height,and establishing a personalized technical database.
作者
刘丽
李政
郭良如
LIU Li;LI Zheng;GUO Liang-ru(College of Sports Science,Jishou University,Jishou 416000,China;School of Sports Science,Hengyang Normal University,Hengyang 421002,China)
出处
《福建体育科技》
2025年第4期84-90,共7页
Fujian Sports Science and Technology
基金
2023年湖南省普通高校教学改革研究项目(项目编号:HNJG 20230872)
2024年湖南省教育科学“十四五”规划项目(项目编号:XJK24CTW013)。
关键词
计算机视觉
羽毛球运动员
羽毛球比赛
跳杀技术
运动学特征
computer vision
badminton players
badminton matches
jump-smash technique
kinematic characteristics
