ECG DENOISING BASED ON EMD FOLLOWED BY MEDIAN FILTER

Abstract

ECG records the electrical activities of the heart and provides the important information which helps in detecting the cardiac abnormalities. But during the ecg acquisition & transmission of signal various type of artifacts or noise such as electromyography (EMG) noise, additive white gaussian noise(AWGN), powerline interferences noise, electrodes contact noise etc get contaminated with the ECG signal and corrupt the main signal. So the ECG signal must have to free from noises for accurate diagnosis of the heart. In this thesis different type of method are used in order to denoised the ECG signal. In this removal of noises is done by various method first by Empirical Mode Decomposition (EMD) followed by Median filter second EMD followed by moving average filter third by Ensemble Empirical Mood Decomposition (EEMD) followed by moving average filter and the last by EEMD followed by median filter. The proposed technique is an enhancement towards the existing EMD and EEMD based denoising algorithm. As ECG is non –stationary signal and EMD is adaptive and data driven so it is suitable for non stationary signal. For the purpose of reducing noisey ECG signal, it is decomposed into Intrinsic mode function(IMF).Using Lower values of IMF, higher frequency value of IMF are neglected before signals being reconstructed and this will be free form noise with higher degree of Signal to Noise Ratio(SNR). Parameter that are used for comparisons are SNR, MSE and Correlation-coefficient.