论文部分内容阅读
胎心音是一种非线性非平稳信号,采集时含有大量噪声,信号处理时需要一种良好的除噪方法。本文先用截止频率为200Hz的巴特沃斯低通滤波器和重采样做预处理,再使用经验模态分解法(EMD)做信号分解,选择含有目标信号的分量进行类小波软阈值自适应算法除噪,最后组合重构得到除噪后的胎心音信号。模态分解时,使用添加掩模信号等方法消除模态混叠,用镜像延拓法消除端点效应,并引用Rilling的研究设定停止规则。该除噪方法一次性地消除了基线漂移和噪声,同小波变换(WT)、数学形态学(MM)和傅里叶变换(FT)比较,信噪比(SNR)明显改善,均方根误差(RMSE)最小,能够满足实际应用的需要。
Fetal heart sound is a non-linear non-stationary signal, collecting a lot of noise, the signal processing requires a good method of removing noise. In this paper, we use the Butterworth low-pass filter with the cut-off frequency of 200Hz and resampling to do the pretreatment, and then use EMD to do the signal decomposition. Select the component containing the target signal to do the wavelet-like soft threshold adaptive algorithm In addition to noise, the final combination of reconstruction to get the noise after the fetal heart sound signal. Modal decomposition, the use of mask signals and other methods to eliminate modal aliasing, mirror extension method to eliminate the endpoint effect, and citing Rilling’s research to set the stop rule. The noise removal method eliminates baseline wander and noise all at once, with a significant improvement in signal-to-noise ratio (SNR) compared with wavelet transform (WT), mathematical morphology (MM) and Fourier transform (FT) (RMSE) minimum, to meet the needs of practical applications.