Recognition of hand motions via surface EMG signal with rough entropy

The rough entropy (RoughEn) is developed based on the rough set theory. It has the advantage of low computational complexity, because there is no parameter to set in RoughEn. In this paper, we characterized the feature of surface electromyography (SEMG) signal with RoughEn and then used support vector machine to classify six different hand motions. The sample entropy, wavelet entropy and approximate entropy were compared with RoughEn to evaluate the performance of characterizing SEMG signals. The experimental results indicated that the RoughEn-based classification outperformed other entropy based methods for recognizing six hand motions from four-channel SEMG signals with the best recognition accuracy of 95.19 ± 2.99%. The results suggest that RoughEn has the potential to be used in the SEMG-based prosthetic control as a method of feature extraction.