基于FPA-VVRKFA的手势表面肌电信号识别  

Surface EMG Signal Recognition of Gestures Based on FPA-VVRKFA

作　　者：季祥 白端元[1] JI Xiang;BAI Duanyuan(School of Electronics and Information Engineering,Changchun University of Science and'Technology,Changchun 130022)

机构地区：[1]长春理工大学电子信息工程学院,长春130022

出　　处：《长春理工大学学报（自然科学版）》2021年第6期109-115,共7页Journal of Changchun University of Science and Technology(Natural Science Edition)

基　　金：吉林省教育厅项目(JJKH20200788KJ)。

摘　　要：表面肌电(Surface Electromyography,sEMG)信号直接、客观地反映了人体肌肉的活动情况,其作为一种便捷的无侵入式肌电检测方法,被广泛地应用于人体动作识别领域。针对表面肌电信号的手势识别问题,提出了一种基于时域特征和向量正则核函数逼近方法(Vector-Valued Regularized Kernel Function Approximation,VVRKFA)的手势识别方法。首先,对MYO臂环采集到的sEMG数据进行活动段检测以提取出活动段;随后,从活动段信号中提取平均绝对值、波形长度、过零点数、均方根和Willison幅值等五个时域特征;最后,应用VVRKFA分类器对提取到的sEMG进行分类识别,同时采用花授粉算法(Flower Pollination Algorithm,FPA)优化分类器参数以保证最佳分类能力。实验结果表明提出的方法在手势动作模式识别上取得了较高的准确率。Surface electromyography(sEMG)signal directly and objectively reflects the activity of human muscles.As a convenient and non-invasive detection method,sEMG was widely used in the field of human motion recognition.Aiming at the problem of hand gesture recognition of sEMG,in this paper,a gesture recognition method was proposed based on timedomain features and vector regularized kernel function approximation(VVRKFA).Firstly,the active segments were detected from the sEMG data collected by MYO arm ring.Then five time-domain features,such as average absolute value,waveform length,zero crossing points,root mean square and Willison amplitude,were extracted from the signals of active segments.Finally,VVRKFA classifier was applied to classify and identify the extracted sEMG;and flower pollination algorithm was used to optimize the classifier parameters to gain the best classification ability.The experimental results show that the proposed method achieves high accuracy in gesture recognition.

关 键 词：表面肌电信号 手势识别 时域特征 向量正则核函数逼近 花授粉算法 

分 类 号：TN911.7[电子电信—通信与信息系统]