摘要
针对时变宽带的呼吸音信号,在分析传统Hilbert变换方法提取包络的缺点基础上,提出基于复小波变换的呼吸音信号包络特征提取方法。选取Morlet复小波,以适当的尺度对预处理后的呼吸音数据进行变换得到包络,提取包络的统计量和能量作为特征,构造BP分类神经网络的输入矢量,经训练识别取得较好分类效果。研究表明该文的特征提取方法是行之有效的。
As the respiratory sound signals are transient and with broad band,an envelope feature extraction method based on complex wavelet transform of respiratory sounds is proposed on the basis of analyzing the shortcomings of Hilbert transform method.Gain the envelope by doing Morlet wavelet transform of respiratory sounds after preprocessing.Extract the statistics and energy of envelope to form the feature vectors of BP neural network.The classification rate is satisfactory.The research findings show that this feature extraction method of respiratory sounds is fairly efficient.
出处
《计算机工程与应用》
CSCD
北大核心
2008年第32期151-154,共4页
Computer Engineering and Applications
关键词
呼吸音
HILBERT变换
复小波变换
包络提取
respiratory sounds
Hilbert transform
complex wavelet transform
envelope extraction
