Myoelectric Gesture Recognition Based on Multi-Channel Correlation Feature

doi:10.3778/j.issn.1002-8331.2111-0123

Abstract

Abstract: The traditional multi-channel surface electromyography（sEMG） gesture recognition extracts the time domain, frequency domain and time frequency domain features of each channel as the input of the classifier. The correlation among channels is rarely considered, and the recognition accuracy is bottlenecked. In order to make full use of the multi-channel sEMG information to improve the accuracy of gesture recognition, a sEMG-based multi-channel correlation feature is proposed. The method calculates the concordance correlation coefficient among multiple channels as the linear correlation feature parameter, and simultaneously obtains the mutual information among the multiple channels as the nonlinear correlation feature parameter. In order to solve the problem of hard estimate of the joint probability density function in mutual information, the method converts mutual information estimate into the estimate of copula entropy, according to the inverse relationship between mutual information and copula entropy. The probability integral transformation is carried out through the empirical distribution function, and non-parametric estimation is adopted to estimate copula entropy, from which the estimate of mutual information is obtained. Then the method constructs the multi-channel correlation feature with the two feature parameters, and inputs the multi-channel correlation feature into the stacking model for recognition. Respectively compared with 4 common time domain features and 5 basic classifiers, the experimental results show that the proposed multi-channel correlation feature can effectively distinguish hand gesture with an average recognition accuracy rate of 94%, demonstrating the effectiveness of the method.

Key words: gesture recognition, surface electromyography（sEMG）, concordance correlation coefficient, mutual information, multi-channel correlation

摘要： 多通道表面肌电信号（surface electromyography，sEMG）传统手势识别方法，主要提取各个通道时域、频域和时频域特征作为分类器的输入，鲜有考虑通道间的相关性，在提升识别精度上遇到瓶颈。为了充分利用sEMG多通道信息以提高手势识别精度，提出一种以多通道相关性为特征的肌电手势识别方法。该方法计算多通道间一致性相关系数，作为多通道sEMG线性相关特征参数，同时获取多通道间的互信息，作为多通道sEMG非线性相关特征参数。实际运用中精确估计联合概率密度函数往往十分困难，根据互信息与copula熵关系，将互信息估计转化为copula熵的估计，通过经验分布函数进行概率积分变换，采用非参数估计方法估计copula熵，从而避免联合概率密度函数的估计。利用两种相关性特征参数构建多通道相关性特征进行对比实验，基于stacking模型使用多通道相关性特征与4种常用时域特征进行识别并对比结果，其次基于多通道相关性特征使用stacking模型与5种常用分类器进行对比识别，实验结果表明所提的多通道相关性特征能有效区分手势动作，在采集的健康受试者手势数据集上平均识别准确率达到94%。

关键词: 手势识别, 表面肌电信号, 一致性相关系数, 互信息, 多通道相关性

JIANG Xi, LI Yanhong, ZOU Ke, YUAN Xuedong. Myoelectric Gesture Recognition Based on Multi-Channel Correlation Feature[J]. Computer Engineering and Applications, 2023, 59(7): 102-109.

江茜, 李沿宏, 邹可, 袁学东. 肌电信号多通道相关性特征手势识别方法[J]. 计算机工程与应用, 2023, 59(7): 102-109.

References

[1] 丁其川，熊安斌，赵新刚，等.基于表面肌电的运动意图识别方法研究及应用综述[J].自动化学报，2016，42（1）：13-25.
DING Q C，XIONG A B，ZHAO X G，et al.A review on researches and applications of sEMG-based motion intent recognition methods[J].Acta Automatica Sinica，2016，42（1）：13-25.
[2] FANG C，HE B，WANG Y，et al.EMG-centered multisensory based technologies for pattern recognition in rehabilitation：state of the art and challenges[J].Biosensors，2020，10（8）：85.
[3] 程淑红，程彦龙，杨镇豪.基于手势多特征融合及优化Multiclass-SVC的手势识别[J].仪器仪表学报，2020，41（6）：225-232.
CHENG S H，CHENG Y L，YANG Z H，et al.Hand gesture recognition based on multi-feature fusion and improved Multiclass-SVC[J].Chinese Journal of Scientific Instrument，2020，41（6）：225-232.
[4] PHINYOMARK A，PHUKPATTARANONT P，LIMSAKUL C.Feature reduction and selection for EMG signal classification[J].Expert Systems with Applications，2012，39（8）：7420-7431.
[5] ENGLEHART K，HUDGINS B.A robust，real-time control scheme for multifunction myoelectric control[J].IEEE Transactions on Biomedical Engineering，2003，50（7）：848-854.
[6] SHI W T，LYU Z J，TANG S T，et al.A bionic hand controlled by hand gesture recognition based on surface EMG signals：a preliminary study[J].Biocybernetics and Biomedical Engineering，2018，38（1）：126-135.
[7] THONGPANJA S，PHINYOMARK A，PHUKPATTARANONT P，et al.Mean and median frequency of EMG signal to determine muscle force based on time-dependent power spectrum[J].Elektronika ir Elektrotechnika，2013，19（3）：51-56.
[8] SRISUWAN N，PHUKPATTARANONT P，LIMSAKUL C.Feature selection for Thai tone classification based on surface EMG[J].Procedia Engineering，2012，32：253-259.
[9] KHUSHABA R N，AL-TIMEMY A H，AL-ANI A，et al.A framework of temporal-spatial descriptors-based feature extraction for improved myoelectric pattern recognition[J].IEEE Transactions on Neural Systems and Rehabilitation Engineering，2017，25（10）：1821-1831.
[10] ENGLEHART K，HUDGIN B，PARKER P A.A wavelet-based continuous classification scheme for multifunction myoelectric control[J].IEEE Transactions on Biomedical Engineering，2001，48（3）：302-311.
[11] PHINYOMARK A，NUIDOD A，PHUKPATTARANONT P，et al.Feature extraction and reduction of wavelet transform coefficients for EMG pattern classification[J].Elektronika ir Elektrotechnika，2012，122（6）：27-32.
[12] 张园园，邹策，陈晓玲，等.基于Gabor小波-传递熵的脑-肌电信号同步耦合分析[J].生物医学工程学杂志，2017，34（6）：850-856.
ZHANG Y Y，ZOU C，CHEN X L，et al.Synchronous analysis of corticomuscular coherence based on Gabor wavelet-transfer entropy[J].Journal of Biomedical Engineering，2017，34（6）：850-856.
[13] STAUDENMANN D，KINGMA I，STEGEMAN D F，et al.Towards optimal multi-channel EMG electrode configurations in muscle force estimation：a high density EMG study[J].Journal of Electromyography and Kinesiology，2005，15（1）：1-11.
[14] FARINA D，JIANG N，REHBAUM H，et al.The extraction of neural information from the surface EMG for the control of upper-limb prostheses：emerging avenues and challenges[J].IEEE Transactions on Neural Systems and Rehabilitation Engineering，2014，22（4）：797-809.
[15] TANG X，LIU Y，LYU C，et al.Hand motion classification using a multi-channel surface electromyography sensor[J].Sensors，2012，12（2）：1130-1147.
[16] MOGK J P M，KEIR P J.Crosstalk in surface electromyography of the proximal forearm during gripping tasks[J].Journal of Electromyography and Kinesiology，2003，13（1）：63-71.
[17] LOWERY M M，STOYKOV N S，KUIKEN T A.A simulation study to examine the use of cross-correlation as an estimate of surface EMG cross talk[J].Journal of Applied Physiology，2003，94（4）：1324-1334.
[18] CHEN X，WANG Z J.Pattern recognition of number gestures based on a wireless surface EMG system[J].Biomedical Signal Processing and Control，2013，8（2）：184-192.
[19] AKBEN S B.Low-cost and easy-to-use grasp classification，using a simple 2-channel surface electromyography（sEMG）[J].Biomedical Research，2017，28（2）：577-582.
[20] 吴亚婷，佘青山，高云园，等.多尺度肌间耦合网络分析[J].生物医学工程学杂志，2021，38（4）：742-752.
WU Y T，SHE Q S，GAO Y Y，et al.Multiple-scale intermuscular coupling network analysis[J].Journal of Biomedical Engineering，2021，38（4）：742-752.
[21] HAN M，REN W，LIU X.Joint mutual information-based input variable selection for multivariate time series modeling[J].Engineering Applications of Artificial Intelligence，2015，37：250-257.
[22] LAWRENCE I，LIN K.A concordance correlation coefficient to evaluate reproducibility[J].Biometrics，1989，45：255-268.
[23] LANGE N，STROTHER S C，ANDERSON J R，et al.Plurality and resemblance in fMRI data analysis[J].NeuroImage，1999，10（3）：282-303.
[24] NOSHAD Z，BOUYERA A，NOSHAD M.Mutual information-based recommender system using autoencoder[J].Applied Soft Computing，2021，109：107547.
[25] 韩敏，刘晓欣.基于Copula熵的互信息估计方法[J].控制理论与应用，2013，30（7）：875-879.
HAN M，LIU X X.Mutual information estimation based on Copula entropy[J].Control Theory & Applications，2013，30（7）：875-879.
[26] KRASKOV A，ST?GBAUER H，GRASSBERGER P.Estimating mutual information[J].Physical Review E，2004，69（6）：066138.
[27] YANG P，YANG Y H，ZHOU B B，et al.A review of ensemble methods in bioinformatics[J].Current Bioinformatics，2010，5（4）：296-308.
[28] DE LUCA C J，GILMORE L D，KUZNETSOV M，et al.Filtering the surface EMG signal：movement artifact and baseline noise contamination[J].Journal of Biomechanics，2010，43（8）：1573-1579.
[29] LAVALLE S M，BRANICKY M S，LINDEMANN S R.On the relationship between classical grid search and probabilistic roadmaps[J].The International Journal of Robotics Research，2004，23（7/8）：673-692.