一种改进阈值的平移不变量小波消噪方法  

A Translation-Invariant Wavelet Denoising Method Based on Improved Threshold

作　　者：涂建成[1] 席旭刚[1] 罗志增[1] 

机构地区：[1]杭州电子科技大学智能控制与机器人研究所,浙江杭州310018

出　　处：《信息与控制》2011年第2期150-154,共5页Information and Control

基　　金：国家自然科学基金资助项目(60874102);国家863计划资助项目(2008AA04Z212);浙江省自然科学基金资助项目(Y1090968)

摘　　要：分析了小波消噪软、硬阈值方法,提出一种改进阈值的平移不变量小波消噪方法.采用重复循环平移原始信号的方法消除不连续点的相互干扰,分别进行小波分解.使用改进的阈值量化算法,通过处理小波系数的幂次和阈值的幂次来估计小波系数,并重构信号,对重构信号进行相反平移,通过相反平移信号求平均得到消噪后的信号、对表面肌电信号处理的实验结果表明,与传统的阈值法相比,该方法提高了信噪比,降低了均方误差.By analyzing the soft threshold and hard threshold,a translation-invariant wavelet denoising method based on improved-threshold is proposed.In order to remove the interference between discontinuous points,the original signal is shifted circulatingly and repeatedly,and then the shifted signals are decomposed by wavelet separately.Afterwards, using the improved threshold quantization algorithm,the wavelet coefficients are estimated by processing the power of the wavelet coefficients and the power of the threshold,and then the signal is reconstructed.The reconstructed signals are shifted reversely,and then the denoised signal is obtained by averaging the reversely-shifted signals.Experimental results of the sEMG（surface electromyography） show that comparing with the traditional threshold methods,this method improves the SNR（signal noise ratio） and reduces the mean square error.

关 键 词：消噪 阈值 平移不变量 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]