Robust hand gesture identification using envelope of HD-sEMG signal

Electromyography (EMG) pattern recognition has been used for different applications such as prosthesis, human-computer interaction, rehabilitation robots, and many industrial applications. In this paper, a robust approach has been proposed for High Density - surface EMG (HD-sEMG) features extraction by using envelopes of HD-sEMG signals. HD-sEMG signals have been recorded by a two-dimensional array of closely spaced electrodes. The recorded signals have been memorized in three datasets of CapgMyo database were employed to ensure the robustness of our experiment. The results display that the spatial features of Histogram Oriented Gradient (HOG) method combined with intensity features have achieved higher performance for Support Vector Machine (SVM) classifier compared with using classical Time-Domain (TD) features for the same classifier.