一种新的HHT在心音包络提取中的应用

摘要

目的:心音包络比原始心音可以更好地显示心音的特征,是进行心音识别的基础.希尔伯特-黄变换(HHT)是一种提取心音包络的有效方法,它首先利用经验模态分解算法提取心音信号的固有模态函数,然后利用希尔伯特变换提取心音包络.常规的希尔波特-黄变换在分解过程中会引起端点效应和过冲等问题.方法:本文提出了一种基于改进型希尔伯特-黄变换的心音包络提取新方法.结果:该方法首先采用包络线性延拓法抑制端点飞翼问题,然后采用三次贝塞尔分段插值算法替代原始经验模态分解算法中的三次样条插值算法减小分解过程中的误差.结论:仿真实验和实际采集的心音信号实验证明了该方法的有效性.%Objective: Envelope of heart sound which can reflect the features of heart sound better than original heart sound do is the basis of recognition of heart sounds. The envelope extraction of heart sounds based on Hilbert-Huang Transform (HHT) is reliable method. Firstly, in order to extract Intrinsic Mode Functions (IMF), the original heart sounds signal is preprocessed by Empirical Mode Decomposition (EMD) algorithm. Secondly, the envelope of heart sounds is got by Hilbert transform. Over-shoots and the end effect become a common phenomenon during the original HHT. Methods: A novel HHT for envelope extraction was presented. Results: Firstly, the approach adopt envelop linearity extension method to solve the problem of end effect. Then, the Cubic Bessel Piecewise interpolation as a substitute for the spine interpolation was used to decrease error during the decomposition. Conclusions: Reliability of this proposed approach is confirmed by computer simulation and real gathering heart sounds signal.