Classification of Surface EMG Signal Based on Energy Spectra Change

In this paper, we introduced a novel and simple methods of extracting the general features of two surface EMG signal patterns: forearm supination (FS) and forearm pronation (FP) surface EMG signals. The surface EMG signal was divided into two segments appropriately at the preparation stage and the action stages. Relative wavelet packet energy (RWPE, symbolized by p_np and p_na respectively at the preparation stage and the action stages, where n denotes the nth frequency band (FB)) of the surface EMG signal was firstly calculated, and then the change (P_n=p_na-p_np) was determined. The results show that P_n from some FBs could effectively characterize the general characteristics of the two surface EMG signal patterns. Compared with P_n in other FBs, P₄ (in 93.75-125 Hz) had more appropriate features.