基于小波变换的心音特征值提取算法研究

介绍了一种新的心音特征值提取算法。利用小波变换进行阈值去噪、包络提取、心音定位、特征值提取,最后识别出心音信号的异常与否。在阈值去噪中利用小波的多分辨率特性对信号进行8层分解并对每层实施软阈值去噪,把去噪后的信号重构,得到去噪后的心音信号,对去噪后的心音信号进行再分解并求出各层香农能量值,踢出包含信息量最少的层后得到心音信号的小波包络,采取自适应阈值线对所得小波包络进行S1,S2定位,并提取特征值T1、T2、T11及T12,利用特征值在二维散点图中的分布情况,实现了心音信号的自识别。

Research on characteristic extraction algorithm of heart sound signal using wavelet transformation

In this work,a new method of heart sound signal self-identification algorithm using wavelet transformation was presented. First, preprocessing heart sound signals were done which including heart sounds denoise with thresholds,extraction of envelope, location of heart sound, and extracting features. Then, identifying the heart sounds were done to test if it was normal or abnormal. In denoise with thresholds of heart sound decomposition 8 layer using wavelet multi-resolution characteristics, and then taking Soft threshold denoising for each layer after decomposition, Now, reconstruction each layer after denoising and get the ideal signal , for this signal taking further decomposition and calculated each layer of Shannon energy value, discard layer that has minimum Shannon energy, and now, we can get envelop of wavelet,adopt the adaptive threshold line proceed this envelop and location S1, S2,last, draw characteristic T1,T2,T11 and T12. Finally, self-identification method was used for heart sound signal using distribution of this characteristic in the two-dimensional scatter plot.