A Fast Classification System for Decoding of Human Hand Configurations Using Multi-Channel sEMG Signals

Abstract

This paper proposes a novel fast classification system consisting of feature extraction and classifier to decode human hand configurations from multi-channel surface electromyogram (sEMG) signals that allows real-time classification of human movement intention as well as prothesis control. In order to enhance the learning speed and the performance of the classifier, we used a supervised feature extraction method (called class-augmented principal component analysis) and a fast learning classifier (called extreme learning machine). Experimental results show that the proposed classification system quickly learns and decodes the human hand configuration with about 92% accuracy.