一种改进的EMD硬阈值去噪算法  

An Improved EMD-based Hard Thresholding Denoising Algorithm

作　　者：张守成[1] 张玉洁[2] 刘海生[1] 

机构地区：[1]华北科技学院基础部,河北廊坊0652012 [2]中国地质大学数学与物理理学院,武汉430074

出　　处：《计算机测量与控制》2014年第11期3659-3661,共3页Computer Measurement &Control

基　　金：国家自然科学基金项目(61071188);中央高校基本科研业务费资助(3142014127);华北科技学院应用数学重点学科资助项目(HKXJZD201402)

摘　　要：为了提升基于经典小波阈值的EMD去噪算法的性能,利用高斯白噪声的统计特征提出了一种改进的硬阈值去噪算法;首先将含噪信号进行EMD分解,把第一个固有模态函数作为高频噪声直接去除并估算出其他IMF中高斯白噪声的能量,然后根据硬阈值去噪的原理,利用滤除掉的样本点包含的能量等于白噪声的能量确定出合适的阈值;该方法能根据样本点自适应地确定阈值;最后通过对含噪正弦信号和仿真心电信号的去噪实验证实了改进后的阈值使算法去噪效果有明显提升。An improved EMD-based hard thresholding denoising algorithm was proposed according to statistical characteristic of white Gaussian noise in order to increase performance of the EMD denoising algorithm based on classical wavelet threshold.Firstly,the noisy signal was decomposed by EMD,the first IMF was removed as high-frequency noise and noise energy included in the other IMFs was estimated.Sec-ondly,threshold parameter was determined based on hard threshold denoising principle and the criterion.that the total energy contained in the removed sample points is equal to the estimated white noise energy.The proposed method is self-adaptive.Finally,the improved method and the original method were imple-mented on noisy sinusoidal signal and artificial ECG signal.The results verify that the presented method can enhance denoising effect obviously.

关 键 词：经验模态分解 固有模态函数 去噪算法 硬阈值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]