A Denoising Algorithm for PPG Signals Incorporating Improved Empirical Modal Decomposition and Wavelet Thresholding

In order to remove multiple noise components contained in PPG signals and extract effective pathological features, this paper combines adaptive full noise ensemble empirical modal decomposition with wavelet thresholding to propose a comprehensive denoising algorithm of PPG for removing multi-component noise contained in PPG. The PPG signal is decomposed to obtain 15 sets of intrinsic modal components (IMF) and residuals (RES), and the correlation coefficients between each component and the original PPG signal are calculated. The method is compared with wavelet thresholding method, EMD joint wavelet thresholding method, EEMD joint wavelet thresholding method and CEEMDAN method, respectively. The experimental results show that the denoising method of CEEMDAN combined with wavelet transform has obvious advantages.