摘要
针对基于表面肌电(sEMG)信号的双臂手势识别率不高的问题,提出一种利用线性判别分析(LDA)方法结合反向传播神经网络(BPNN)算法的手势识别方法。首先,对采集的双臂sEMG信号进行小波阈值去噪的预处理,提取信号中的均方根值、绝对值均值、过零点次数、立方均值、波长、平均绝对值斜率共6种特征;再通过LDA对高维特征集进行降维处理;最后,利用BPNN建立相应的手势模型并识别。实验结果表明:在双臂手势动作的背景下,该识别算法效率较高,识别准确率高达92.7%,能够有效实现双臂手势识别。
Aiming at the problem of low rate of double-arm gestures recognition based on surface electromyography(sEMG)signal,a gesture recognition method using linear discriminant analysis(LDA)method and back propagation neural network(BPNN)algorithm is proposed.Firstly,wavelet threshold denoising preprocessing on double-arm sEMG is carried out,and extract six kinds of features of root mean square value of the signal,the mean absolute value,the number of passing zero,the mean of cubic value,wavelength,the average absolute value slope,by LDA dimension reduction processing high dimensional feature set is carried out.Finally,corresponding gesture model is established based on BPNN and identified.Experimental results show that the algorithm has high efficiency and the recognition accuracy is as high as 92.7%under the background of double-arm gesture actions,which can effectively realize the recognition of double-arm gesture.
作者
王金玮
曹乐
阚秀
张文艳
孟壮壮
WANG Jinwei;CAO Le;KAN Xiu;ZHANG Wenyan;MENG Zhuangzhuang(School of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《传感器与微系统》
CSCD
北大核心
2023年第6期158-160,168,共4页
Transducer and Microsystem Technologies
关键词
表面肌电信号
小波阈值去噪
线性判别分析方法
反向传播神经网络
手势识别
surface electromyograph(sEMG)signal
wavelet threshold denoising
linear discriminant analysis(LDA)method
back propagation neural network(BPNN)
gesture recognition
