Performance Analysis of Adaptive Filters with various Wavelets for Noise Removal in EEG Signals

Abstract

Noise removal from recorded EEG signal is most essential for better analysis of brain disorders. During recoding time, EEG signals are usually contaminated by various noise and distortions due to several artifacts. These noisy EEG signals may lead to wrong diagnosis of brain disorders. There are several techniques available to remove the noise from EEG signals. But these techniques are unable to remove the noise completely. However, they can minimize the noise in EEG signals so that the physicians can predict brain disorders. This work presents to minimize the noise by Discrete Wavelet Transform Methods using haar, db2, symlet and coiflet wavelets. EEG original signals from public EEG database are used for experimentation and wavelet transformations, are applied by using Matlab code. The filters performance is measured and analyzed on the basis of performance parameters like SNR and MSE which are calculated for various step sizes of signal and filter orders. Wavelet analysis techniques shows better performance when compared to others