NNMF Analysis to Individual Identification of Fingers Movements Using Force Feedback and HD-EMG

Abstract

High-density electromyography (HD-EMG) signals have been widely used today due to their ability to extract spatial information from muscle activity. Several studies pursuit a pattern of activation of the fingers with the muscle activation center using several types of different protocols. This work aims to develop a feedback protocol for finger recognition using HD-EMG. The tools for the analysis were the activation centroids of each finger in addition to the use of the non-negative matrix factorization (NNMF) algorithm to extract the associated synergies of each finger and dissociate them. The protocol elected in this research proved to be effective in the selection of index and minimum fingers. In addition, an ANOVA statistical test application on synergies showed great utility in dissociating fingers.