摘要
提出了一种新的ECG信号降噪方法.该方法以Hilbert-Huang变换(HHT)和阈值降噪方法为核心,将经验模态分解(EMD)分解出的各层固有模态函数(IMF)分为噪声成分起主导作用层和有用信号起主导作用层.对噪声成分起主导作用层采用Donoho等人提出的阈值估计法,对有用信号起主导作用层的噪声水平通过高频层的能量与平均周期的乘积来确定.讨论了噪声成分起主导作用层和有用信号起主导作用层的区分判别问题.该方法克服了阈值估计法对有用信号起主导作用层噪声水平估计的较大偏差.经实验验证表明,该方法能有效地滤除ECG信号检测中的几类主要噪声,且失真很小.
A new ECG signal noise reduction method is presented. The core of this method is Hilbert-Huang transform （HHT） and threshold denoising. It divides intrinsic mode functions （IMF） that decomposed using empirical mode decomposition （EMD） into noise dominative layers and useful signal dominative layers. Threshold denoising proposed by Donoho et al. is adapted for the noise dominative layer, and for useful signal dominative layer. The noise level can be obtained by the product of the energy and mean period of the high frequency layers. The problem how to distinguish the noise dominative layer and useful signal dominative layer was discussed. This method overcomes estimated level deviation of the noise for useful signal dominative layer using threshold denoising. Experiments indicate that several main noise of ECG signal detecting can be removed well and the distortion is very little.
出处
《传感技术学报》
CAS
CSCD
北大核心
2006年第6期2578-2581,2590,共5页
Chinese Journal of Sensors and Actuators
基金
黑龙江省教育厅科学技术研究项目资助(11511086)
作者简介
宋立新（1963-），男，教授，博士研究生，研究方向为弱信号检测与信号处理，lixins99@yahoo．com．cn；
王祁（1944-），男，教授，博士生导师．
