CEEMDAN-小波包联合算法在ECG中的降噪应用

Application of Noise Reduction of ECG Based on CEEMDAN-Wavelet Packet Combined Algorithm

针对心电信号采集过程中存在的多种噪声干扰,提出了一种完备的自适应噪声经验模态分解(CEEMDAN)和小波包联合去噪算法。该方法首先通过CEEMDAN分解含噪心电信号得到一组固有模态分量函数(IMF),之后计算每个IMF与原心电信号的相关系数,绘制出IMF分量的频谱图。将相关系数和IMF频谱图相结合来筛选出含噪明显的本征模态分量,对高频含噪分量进行小波包阈值去噪处理,对低频含有基线漂移的分量用中值滤波器滤除,最后将信号重构得到去噪后的信号。以MIT-BIH心律失常数据库的数据为实验样本,分别应用现有的四种算法和该算法做去噪处理,结果表明经该算法处理过的信号信噪比更高,均方根误差更小,降噪效果更佳。

A Complete Ensemble Empirical Mode Decomposition with Adaptive Noise Analysis(CEEMDAN)and wavelet packet joint denoising algorithm is proposed to solve the various noise interference in ECG signal acquisition.In this method,a set of intrinsic modal functions(IMF)are obtained by CEEMDAN decomposition of noisy ECG signals,and then the correlation coefficients between each IMF and the original ECG signals are calculated to draw the spectrum diagram of IMF components.The correlation coefficients are combined with the IMF spectrum to filtrate the eigenmode components with obvious noise.The high-frequency components with noise are denoised by wavelet packet threshold,and the low-frequency components with baseline drift are filtered by median filter.Finally,the signal is reconstructed and the denoised signal is obtained.Taking the data of MIT-BIH arrhythmia database as the experimental samples,the existing four algorithms and the algorithm are respectively applied for denoising processing.The results show that the signal-noise ratio processed by the algorithm is higher,the root mean square error is smaller and the denoising effect is better.