Finger Position Classification from Myoelectric Signal using Time Domain Based Features

Abstract

In this paper the classification of finger gesture using multichannel surface electromyography (sEMG) signals is proposed. Three types of hand gestures were applied to be identified, when participants holding spherical objects. A specific finger position is evaluated while the hand grasps a specific geometrical object, whose shape is considered as a way to introduce a parameter variation. Hand natural motions are collected by placing electrodes on five muscles on the forearm of six healthy subjects who are fastenings spheres into a controlled environment. A feature vector approach in time domain (TD) is given as input to a linear discriminant analysis (LDA) module used as statistical pattern classifier. We show that it is possible to categorize each motion, that is, TD feature based provide an effective representation for classification, indicating the membership of myoelectric signals (MES) collected to a finger positions class. These results will be useful for human hand motion analysis and has potential applications especially in robotic hand or prosthetic hand control and human--computer interaction (HCI).