Classifying of EEG Signals Recorded During Right and Left-hand Finger Movements

Abstract

Brain Computer Interface (BCI) technology allows a person to control a device by bypassing the use of muscular activity. Signal processing and classification methods play a decisive role in the performance accuracy in BCI application. In this thesis extensive comparison among novel electroencephalic(EEG) pattern recognition methods is provided. Signals collected during left/right self-paced typing are analyzed and classified based on different schemes including Autoregressive and Exogenous Autoregressive model estimation, Smoothing and Time Averaging and Common Spatial Patterns (CSP) filtering. Comparison between methods is performed mainly on the BCI 2002 and 2003 competition data sets available on the Internet and currently used by many researchers as etalon data sets. The proposed methods combining common spatial pattern filtering feature extraction and Mahalanobis distance classifier as well as Support Vector Machines show the best performance.