Different scenarios on denoising of signals in the ıntrinsic mode function selection framework

Abstract

Without having any information of original signal, estimating the desired signal from noisy measurements is a challenging problem. In this paper, the denoising problem of signals corrupted by additive white Gaussian noise (AWGN) is considered in the empirical mode decomposition (EMD) framework, and five different noise suppression scenarios based on the various combinations of intrinsic mode functions (IMFs) that arise from applying the EMD to a given noisy signal are suggested. In these scenarios, the idea of discarding noise-dominant IMFs from a noisy signal is adopted. Considering the root-mean-square error and the signal-to-error ratio, the performance of each scenario is evaluated over simulated and real signals contaminated by AWGN with different signal-to-noise ratios (SNRs). It is observed from simulations that the proposed scenarios provide satisfactory denoising performance especially for positive SNRs and can be exploited as a primary stage in whole of the noise-diminishing applications. © 2016 IETE.