Enhancing classification accuracy of wrist movement by denoising sEMG signals

This paper presents identification of 4 different wrist movements by analyzing fore-arm surface Electromyogram (sEMG) signals. In order to reduce noise picked up during the recording, wavelet based denoising is applied using Daubechies mother wavelet. Spectral features along with Wilson´s amplitude were extracted and given to a linear classifier. The experimental result shows better recognition performance using the given features when denoising is applied. The maximum accuracy for identification of four wrist movement was 97.5% which is quite significant as compared to the previous researches.