SEMG De-Noising Based on the Lifting Wavelet Transform

In order to improve the surface electromyography (SEMG) pattern recognition ability of hand movement, this paper presents a de-noising method based on lifting wavelet transform. Firstly, high frequency detail coefficients of multilayer signals are obtained from original SEMG using the lifting wavelet decomposition with lifting algorithm. Then the coefficients are treated by the soft and the hard threshold method separately. Finally, a noise decreased signal is obtained through reconstructing the filtered coefficients. The de-noising experiments of standard sine adding noise signal and real SEMG are carried on. The results show that the lifting wavelet is an obvious better de-noising method compared to the first generation wavelet. In addition, the hard threshold method is more ideal for SEMG de-noising than the soft threshold method.