基于小波变换的心音信号滤波方法研究  

Research of De-noising Method of Heart Sound Signals Based on Wavelet Transform

作　　者：陈天华[1] 韩力群[1] 邢素霞[1] 郭培源[1] 

机构地区：[1]北京工商大学计算机与信息工程学院,北京100048

出　　处：《计算机仿真》2010年第12期401-405,共5页Computer Simulation

摘　　要：为了准确提取心音信号,心音信号消噪是实现心血管疾病无创诊断的前提,传统的心音信号消噪方法在消除噪声的同时,也滤除了心音的有用信息。利用小波变换多尺度多分辨率的特点对心音信号进行分解,将不同频带的信号展现在小波分解的不同尺度上,根据心音信号的频率分布特点,通过选择重构系数可消除心音信号中的干扰分量,并对不同小波的消噪效果以及同一小波不同分解层数的消噪效果进行仿真对比研究。结果表明,在haar、db6、sym8、coif5四种常用小波中,db6小波对心音信号的消噪效果最明显,证明其中采用db6小波进行5层分解可以获得最佳消噪效果,并对临床采集的228例心音信号进行消噪实验,db6小波均获得了很好的效果。De-noising of heart sound signals is the primary basis for achieving non-invasive diagnosis of coronary heart disease,the traditional heart sound signal de-noising method to eliminate noise while also filtered heart sound of effective information.This paper aims to study the de-noising methods of suited to the characteristic of heart sound signals.In this article,according to the multidimension and multiresolution of wavelet transform,heart sound signals at different frequencies are decomposed using multidimensional wavelet transform,and according to the frequency distribution features of heart sound signal,various levels of de-noising of heart sound signals are achieved through the reconstruction of wavelet coefficients.We have done a research of comparing the de-noising effect of different wavelets and de-noising effect of the same wavelet at different levels.As our series of experiments have shown,of the four most commonly used wavelets,i.e.db6,haar,sym8,and coif5,db6 is capable of producing the best denoising effect;while the same wavelet at different levels yields different denoising effects,level-5 db6 decomposition accomplishes the most desirable result.These experiments have been conducted upon the heart sound signals of a clinical sample of 228 people,db6 achieving an ideal result.

关 键 词：心音信号 小波变换 消噪 无创诊断 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]