DWT-Based Hand Movement Identification of EMG Signals Using SVM

Electromyogram (EMG) signals are widely used in rehabilitation, medical, engineering, robotic, and industrial fields. For amputee’s residual muscles control, EMG signals have been utilized. Multiple-level approximation and detail coefficients of EMG signals were studied here. In this study, EMG signals were extracted from 20 healthy human subjects by multilevel coefficients of wavelet. After the EMG data acquisition, preprocessing was done followed by discrete wavelet transform (DWT) de-noising and feature extraction by db4 wavelet. Three-level decomposition was achieved with DWT technique, and higher classification accuracy was observed by support vector machine (SVM) classifier with principal component analysis. This paper proposed a combination of a highly accurate algorithm for the pattern recognition of EMG signal with different movements. The finding of this paper could be utilized with real-time applications.