sEMG신호기반 특징 및 채널 선택을 이용한 다중 보행 단계 인식

초록

In this paper, we propose a sEMG signal based five-step gait phase recognition method using feature and channel selection. We observed that EMG signal patterns may vary according to subject’s locomotion convention. The proposed method can enhance the classification accuracy by selecting proper channels and features according to recognition rate ranking of channels and features for each subject in training phase. Experimental results show that the classification accuracy of the method using small number of channels and/or features is better than that using large number of channels and/or features.