基于经验模态分解-小波包变换的表面肌电信号手势识别

为了提高表面肌电信号（sEMG）手部运动识别的正确率，比较常规的sEMG预处理和特征提取方法，提出一种基于经验模态分解（EMD）和小波包变换（WPT）的sEMG手势识别模型。首先，使用EMD方法将sEMG进行平稳化，得到一系列的固有模态函数。其次，求取各个固有模态函数与原始信号的相关性，选取相关性较高的前4个分量作为有效分量。然后，采用Db3小波函数进行WPT，提取小波包系数中的平均能量、平均绝对值、最大值、均方根和方差等特征。分别采用线性判别分析和支持向量机对12种手部运动进行模式识别。结果表明基于EMD和WPT的sEMG手势识别正确率比直接提取小波包系数中的特征识别正确率高。

Empirical mode decomposition and wavelet packet transform applied to surface EMG signal for hand gesture recognition

Abstract: Compared with conventional surface electromyography (EMG) signal preprocessing and feature extraction methods, a model based on empirical mode decomposition and wavelet packet transform for hand gesture recognition through surface EMG signals is proposed for improving the accuracy of surface EMG signal in hand gesture recognition. Empirical mode decomposition is firstly used to smooth the surface EMG signal for obtaining a series of intrinsic mode functions. Subsequently, the correlation between each intrinsic mode function and the original signal is obtained, and the top 4 components with higher correlation are selected as effective components. Then, Db3 wavelet function is used to perform wavelet packet transformation, and the average energy, average absolute value, maximum value, root mean square and variance of the wavelet packet coefficients are extracted. Finally, linear discriminant analysis and support vector machine are used to recognize 12 hand gestures separately. The results show that the hand gesture recognition accuracy of applying empirical mode decomposition and wavelet packet transform to surface EMG signal is higher than that of directly extracting wavelet packet coefficients.