Empirical Copula driven hand motion recognition via surface electromyography based templates

Abstract

Current tendency of electromyography (EMG) based prosthetic hand is to enable the user to perform complex grasps or manipulations with natural muscle movements. In this paper, a new classifier is introduced to identify the naturally contracted surface EMG patterns for hand motion recognition. The recognition method utilizes a dependence structure as a motion template, which includes one-to-one correlations of surface EMG feature channels. Using an effective EMG feature, the proposed algorithm can successfully classify different complex motions from different subjects with a satisfactory recognition rate. To save the computational cost, re-sampling processing has been employed. © 2010 Springer-Verlag.