基于组合能量特征的表面肌电信号手势识别算法  

Surface EMG signal hand motion recognition algorithm based on combinated energy characteristics

作　　者：宋佳强 裴晓敏[1] 赵强[1] 刘洪海 SONG Jiaqiang;PEI Xiaomin;ZHAO Qiang;LIU Honghai(School of Information and Control Engineering,Liaoning Shihua University,Fushun 113001,China;Intelligent Systems and Biomedical Robotics Group,University of Portsmouth,PO13QL,Portsmouth)

机构地区：[1]辽宁石油化工大学信息与控制工程学院,辽宁抚顺113001 [2]朴茨茅斯大学智能系统与生物医学机器人实验室,英国朴茨茅斯PO13QL

出　　处：《传感器与微系统》2020年第6期139-142,共4页Transducer and Microsystem Technologies

基　　金：辽宁省自然科学基金资助项目(20180551056)。

摘　　要：为了提高基于表面肌电信号(sEMG)的手势动作识别准确率,提出一种基于肌电信号组合能量特征的手势识别方法。首先,计算s EMG信号和其高阶差分信号的能量谱;然后,提取基于能量谱的组合特征;最后,用主成分分析(PCA)算法对多组能量特征维度约简,线性判决分析(LDA)分类器识别手势动作。利用肌电仪采集8组手势动作进行识别,基于组合能量特征的肌电信号手势识别方法对于手势动作识别的准确率可达97. 5%,比基于典型特征提取算法手势动作识别准确率更高;利用UCI数据库中的肌电信号进行实验,手势识别准确率可达94. 5%。实验表明:组合能量特征提取算法对不同的数据库具有普适性,所提取肌电信号组合能量特征能使不同手势动作的差异性更明显,整个手势识别方法准确率更高。In order to improve the accuracy of gesture recognition based on surface electromyography signal( s EMG),a gesture recognition method based on the combined energy characteristics of electromyography signal is proposed. Firstly,the energy spectrum of s EMG signal and its high-order differential signal are calculated. Then,the combined features based on energy spectrum are extracted. Finally,PCA algorithm is used to reduce the multigroup energy feature dimension,and the LDA classifier is used to recognize gestures. Eight groups of gestures are collected by electromyography for recognition,and the recognition accuracy of electromyography signal gesture recognition method based on combined energy features can reach 97. 5%. which is higher than that based on typical feature extraction algorithm. By using the EMG signal in UCI database,the accuracy of gesture recognition can reach 94. 5 %. Experiments show that the combined energy feature extraction algorithm is applicable to different databases,and the extracted EMG combined energy feature can make the difference between different gestures more obvious,and the accuracy of the whole gesture recognition method is higher.

关 键 词：表面肌电 特征提取 手势识别 降维 线性判别式分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP212[自动化与计算机技术—计算机科学与技术]