EMG Based Classification of Hand Gesture Using PCA and SVM

Abstract

Biomechanics is a field of science that studies the movement of living things, especially humans. Bio-mechanical science produces new technology, namely electromyography (EMG). Electromyography (EMG) is a technique for recording signals originating from human muscles during contraction or relaxation, so it is widely used as a control medium. One of the applications of the EMG is in controlling the robotic arm. To do this task the recognizing the signal of the EMG for hand gestures is needed. This study aims to identify five finger movement patterns using a Myo Armband sensor based on electromyography (EMG) signals. This equipment is located on the forearm of the user’s right hand to get a signal from the EMG. The 70 percent EMG signal is used as training to get the weight of the results for each movement. Then, the weight of the training results is tested using 30% of the EMG signal data and grouped using the Support Vector Machine (SVM) method. At the classification stage, a success percentage of 60% was obtained for sensor 3, 73% for sensor 4, and 53% for sensor 8. Furthermore, after receiving a percentage of the success of this study, the authors hope this research can become a reference for the development of making hand robots for medical purposes.