基于CEEMDAN与自相关函数的心音去噪
Heart Sound Denoising Based on CEEMDAN and Autocorrelation Function
唐瑭 1卢官明 1戚继荣 2王洋 2赵宇航2
作者信息
- 1. 南京邮电大学通信与信息工程学院,江苏 南京 210003
- 2. 南京医科大学儿科学院,江苏 南京 210008
- 折叠
摘要
为有效去除心音信号中的噪声,提出基于 CEEMDAN(Complete Ensemble Empirical Mode Decomposition with Adaptive Noise)与自相关函数的心音去噪算法.首先,通过CEEMDAN将含噪的心音信号分解为具有不同尺度特征的IMF(Intrinsic Mode Function)分量;其次,根据噪声与心音的自相关函数性质不同,界定IMF分量的信噪分界点;最后,对以噪声为主的IMF分量进行均值滤波,并将其与以心音为主的IMF分量重构得到去噪后信号.实验表明,在不同的噪声水平下,与小波软阈值去噪算法、小波硬阈值去噪算法、CEEMDAN去噪算法相比,所提算法的信噪比最高,均方根误差最小,在去除噪声的同时,可以较好地保留心音信号中的有效信息.
Abstract
To effectively remove noise from heart sound signals,a heart sound denoising algorithm based on CEEMDAN(Complete Ensemble Empirical Mode Decomposition with Adaptive Noise)and the autocorrelation function is proposed.First,the noisy heart sound signal is decomposed into Intrinsic Mode Function(IMF)components with different scale features through CEEMDAN.Next,the difference in properties between the noise and the heart sound autocorrelation functions is used to define the signal-to-noise threshold for the IMF components.Finally,mean filtering is applied to the IMF components dominated by noise,and the denoised signal is reconstructed by combining these with the IMF components dominated by heart sound.Experiments show that,at various noise levels,the proposed algorithm achieves the highest Signal-to-Noise Ratio(SNR)and the lowest Root Mean Square Error(RMSE)compared to wavelet soft and hard threshold denoising algorithms and the CEEMDAN denoising algorithm.It effectively removes noise while preserving the essential information in the heart sound signal.
关键词
心音去噪/CEEMDAN/自相关函数/均值滤波Key words
heart sound denoising/CEEMDAN/autocorrelation function/mean filtering引用本文复制引用
出版年
2025
NETL
NSTL
万方数据