Pre-processing of EEG signal using Independent Component Analysis

Biometric systems play vital role in everyday life. After signal acquisition, first step for any biometric system is to process the signal and extract valuable information from that signal. Pre-processing the EEG signal, which is essentially a collection of signal processing techniques before the primary EEG data analysis, is required to extract just brain activity from noisy EEG data. The selection of pre-processing techniques can influence the results of future EEG data analysis. Pre-processing of EEG signal includes various techniques like reduction of line noise, removal of redundant channels and reducing artefacts from EEG signal. EEG data is highly contaminated from various artefacts. This paper implements ICA as a pre-processing technique to remove eyeblink artefacts from signals which is then followed by Auto Regressive model features and results are obtained using Support Vector Machine classifier. Results shows that EEG signals performs better after applying ICA to raw data.