基于动作识别算法的健美操难度自动评分系统设计  

Design of Automatic Scoring System of Aerobic Difficulty Based on Action Recognition Algorithm

作　　者：马倩倩 贺莉 MA Qianqian;HE Li(School of Physical Education,West Anhui University,Lu′an,Anhui 237012,China)

机构地区：[1]皖西学院体育学院,安徽六安237012

出　　处：《西昌学院学报（自然科学版）》2021年第2期106-110,共5页Journal of Xichang University(Natural Science Edition)

基　　金：安徽省体育社会科学研究项目(ASS2016203)。

摘　　要：为了提高健美操评分的精度,减少人为主观性评价的影响,设计基于动作识别算法的健美操难度自动评分系统。系统数据采集层利用体感传感器Kinect采集健美操运动员体感信息,输出健美操运动员动作图像;将所获取健美操动作图像使用串口通信协议传输至数据处理层;数据处理层使用基于最大相关-最小冗余的动作识别算法,识别健美操动作后传输至应用层;应用层按照健美操难度评分标准自动评价健美操难度。实验结果显示:所设计系统识别健美操动作时,识别结果的均方误差、平均绝对误差较小,识别精度较高;对10位运动员健美操动作难度系数评分值与实际难度系数的差值较小,评分结果精度较高,可以为健美操竞技评分提供依据。In order to improve the accuracy of aerobics scoring and reduce the influence of subjective evaluation,an automatic scoring system of aerobics difficulty based on action recognition algorithm is designed.The data acquisition layer of the system collects the aerobic athletes’somatosensory information and outputs the athletes’action images through Kinect;the obtained aerobics action images are transmitted to the data processing layer with serial communication protocol;the data processing layer uses the action recognition algorithm based on maximum correlation and minimum redundancy to recognize the aerobics actions and then transmit them to the application layer;the application layer automatically scores the aerobic difficulty according to the aerobic difficulty criterion.The experimental results show that:when the designed system identifies aerobic movements,the mean square error and the mean absolute error of the recognition results are relatively small,and the recognition accuracy is high;the difference between the difficulty coefficient scoring results and the actual difficulty coefficients of 10 athletes’aerobic movements is small,and the scoring accuracy is high.Therefore,it can provide reference for aerobic competition scoring.

关 键 词：动作识别 健美操难度 自动评分系统 最大相关 最小冗余 体感传感器 

分 类 号：G831.3[文化科学—体育训练] TP391.41[文化科学—体育学]