Hand Gesture Recognition using sEMG Signals Based on Support Vector Machine

Abstract

This paper presents a method based on support vector machine (SVM) to recognize hand gestures using surface electromyography (sEMG). In our method, Myo armband (a sEMG device with eight channels) is used to measure subjects' forearm sEMG signals. The original sEMG signal is preprocessed to reduce noise and detect muscle activity regions. Feature extraction is applied by segmenting a sliding sub-window in the preprocessed signals to get each segment of signals. Connect the signal segment with the results using a bag of functions to generate a feature vector. For classification, we train a SVM classification model which includes 5 sub-models. Each sub-model can recognize a gesture, like fist, wave in, wave out, fingers spread, and double pinch. Finally, we test the proposed model to recognize these gestures, and achieve an accuracy of 89.0%.