基于小波变换的呼吸音降噪新方法研究.docVIP

基于小波变换的呼吸音降噪新方法研究 摘 要： 为了对临床采集的含噪声呼吸音信号进行降噪，探讨了基于小波多分辨率分解和重构的两种方法：基于小波的自适应阈值（AWT）降噪方法和基于平稳?非平稳滤波技术（ST?NST）降噪方法。AWT将呼吸音作为噪声先行去除，通过参数调解，对采集信号每一层高频小波系数自适应的进行阈值量化；ST?NST通过平稳?非平稳滤波技术将呼吸音和心音分离到两个空间，分别通过重构两个空间的小波系数，重构出呼吸音。通过对标准信号及对临床采集呼吸音提取实验，ST?NST对正常人支气管呼吸音降噪、AWT对哮喘病人哮鸣音降噪分别都有很好的效果。以上两种方法，具有较强的实用价值，可以得到高信噪比的呼吸音信号，为后续呼吸音特征提取和分类提供了基础。 关键词： 呼吸音； 心音； 小波自适应阈值； 平稳?非平稳滤波技术； 降噪 中图分类号： TN911.72?34 文献标识码： A 文章编号： 1004?373X（2016）03?0018?05 Research on wavelet transform based new denoising methods for respiratory sound QIN Guojin1， WU Zhaoping2， WANG Xinping2， FANG Yu1， WANG Haibin1， GAN Fengping1 （1. College of Electrical Engineering and Electronic Information， Xihua University， Chengdu 610039， China； 2. Medicine of Respiration， The No. 452 Hospital of PLA， Chengdu 610021， China） Abstract： To denoise the noise?involved respiratory sound signal， which is collected from the clinic， two denoising methods based on the wavelet multi?resolution decomposition and reconstruction are discussed， one is based on the adaptive wavelet threshold （AWT）， and another is based on the stationary?non?stationary filtering technology. The respiratory sound as the noise is eliminated with the former method， and then the threshold quantization for the high?frequency wavelet coefficient of each layer of the acquisition signal is conducted by parameter mediation. The respiratory sound and heart sound are separated into two spaces by the stationary?non?stationary filtering technology， and then the respiratory sound is reconstructed by reconstructing the wavelet coefficients of the two spaces. The extraction experiments of standard signals and respiratory sound collected from the clinic were conducted. The experiment results show that the method based on the stationary?non?stationary filtering technology has good effect on bronchial respiratory sound denoising for the normal patients， and the method based on AWT has good effect on wheezing denoising for the asthma patients. The two methods have strong practic