Surface electromyography feature extraction based on wavelet transform

Considering the vast variety of EMG signal applications such as rehabilitation of people suffering from some mobility limitations, scientists have done much research on EMG control system. In this regard, feature extraction of EMG signal has been highly valued as a significant technique to extract the desired information of EMG signal and remove unnecessary parts. In this study, Wavelet Transform (WT) has been applied as the main technique to extract Surface EMG (SEMG) features because WT is consistent with the nature of EMG as a nonstationary signal. Furthermore, two evaluation criteria, namely, RES index (the ratio of a Euclidean distance to a standard deviation) and scatter plot are recruited to investigate the efficiency of wavelet feature extraction. The results illustrated an improvement in class separability of hand movements in feature space. Accordingly, it has been shown that only the SEMG features extracted from first and second level of WT decomposition by second order of Daubechies family (db2) yielded the best class separability