优化VMD与NLM结合的信号去噪

Signal denoising based on optimized VMD and NLM

为解决人体心电信号易出现噪声干扰的情况,提出一种自选取参数的变分模态分解(variational mode decomposition)与非局部均值(non-local means)ECG信号去噪方法。针对人为选取变分模态分解参数不准确的问题,运用一种自适应的参数确定方法,根据不同的原始信号,确定不同的参数并对信号进行分解,得到一组信号分量,计算各个分量的样本熵,根据样本熵的值,选取出噪声主导分量和有效分量,对噪声主导信号进行NLM去噪,将去噪后的信号分量与剩余的有效信号分量进行重构得到去噪信号。实验结果表明,改进参数的VMD-NLM算法相较其它算法,去噪信号的信噪比(SNR)和均方误差(MSE)最好,对人体心电信号的去噪效果最为优秀。

To solve the situation that the human ECG signal is prone to noise interference,a variational mode decomposition(VMD)based on self-selected parameters and non-local mean(NLM)signal denoising method was proposed.In response to the problem of artificially selecting the variational modal decomposition parameters,an adaptive parameter determination method was used.According to the different original signals,the different parameters were determined,the signals were decomposed to obtain a set of signal components and the sample entropy of each component was calculated.Based on the sample entropy,the noise dominant signal and effective signal were distinguished,the noise dominant signal was denoised through NLM.The denoised signal and the effective signal were reconstructed to obtain a reconstructed signal.The results show that comparing with other algorithms,the VMD-NLM algorithm with improved parameters has the best signal noise ratio(SNR)and mean square error(MSE)of the denoised signal,and the best denoising effect on the human ECG signal.