基于惯性传感的越野滑雪技术动作识别  

Cross-country skiing technique action recognition based on inertial sensing

作　　者：张振鹏 杨云 陈亮 何剑 ZHANG Zhenpeng;YANG Yun;CHEN Liang;HE Jian(School of Instrument and Electronics,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学仪器与电子学院,山西太原030051

出　　处：《传感器与微系统》2022年第12期126-129,共4页Transducer and Microsystem Technologies

基　　金：国家重点研发计划资助项目(2019YFB2004802,2019YFF0301802,2018YFF0300605)。

摘　　要：越野滑雪技术动作识别能够有效帮助运动员优化滑行策略和改进技术动作。传统方法存在空间受限、变更装备及传感器数目较多的问题,因此,提出了一种基于惯性传感的越野滑雪动作识别方法。对惯性数据进行低通滤波的预处理,实现越野滑雪子技术动作区间划分。生成时域频域特征并对其特征值进行归一化和主成分分析(PCA)。设计了支持向量机(SVM)分类器,实现对越野滑雪技术动作的识别。实验结果表明:该方法对3种越野滑雪子动作双杖推撑(DP)滑行,对角跨步(DS),踢腿双杖推撑(KDP)的识别率高达95.4%,证明了所提方法为越野滑雪动作识别提供了一种新颖高效的解决方案。Cross-country skiing technical action recognition can effectively help athletes optimize ski strategies and improve technical actions.Traditional method has the problems of limited space, changing equipment and a large number of sensors.Therefore, a cross-country skiing action recognition method based on inertial sensing is proposed.The inertial data is preprocessed by low-pass filtering to realize the subaction interval division of cross-country skiing.Time-domain and frequency-domain features are generated, and normalization and principal component analysis(PCA)on the feature values are performed.Support vector machine(SVM)classifier is designed to realize the recognition of cross-country skiing technical actions.The experimental results show that the recognition rate of the method on three cross-country skiing sub-actions DP,DS and KDP is as high as 95.4 %,which proves that the method provides a novel and efficient solution for cross-country skiing action recognition.

关 键 词：越野滑雪 动作识别 惯性测量单元 支持向量机 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] G863.13[自动化与计算机技术—控制科学与工程]