中文摘要：

根据在不同尺度下信号和噪声的小波变换系数的相反特性，提出了一种改进的小波消噪算法来去除肌电信号中的噪声．利用Mallat算法对肌电信号进行小波分解，实质上就是将信号投影到尺度空间和小波空间，分别包含了信号的光滑通道分量和细节分量．兼顾软阈值和硬阈值量化方法的优点，利用两者的加权平均值滤除由噪声所决定的小波变换系数，从而在大尺度下补充细节信息并保持信号在奇异点的特征．利用保留下来的小波变换系数进行信号重构即得到消噪后的信号．实验结果表明，该方法可以有效去除噪声，兼顾了软、硬阈值的优点，保留了在模式变化过程中肌电信号细节部分的有用信息．

英文摘要：

According to the inverse characteristics of useful signal and noise in different wavelet scales, a novel wavelet denoising method was proposed to eliminate the noise in electromyography （EMG） signal. Wavelet decomposition using Mallat arithmetic projected the signal to scale space and wavelet space in essence, which included the smooth channel and the detail weight respectively. To reserve the advantages of soft-threshold and hard-threshold methods, the mean value of the two thresholds was used to remove the wavelet coefficients caused by the noisiness, so detailed information was supplied in large scales and the signal singular feature was reserved at the same time, Finally, using the reserved information, the denoised signal was obtained based on the wavelet reconstruction algorithm. Experimental results show that the method has good noise removing performance, reserves the advantages of soft-threshold and hard-threshold methods, and holds the useful detailed EMG information during pattern changes.