DefinePK

Abstract

A Brain-Computer Interface (BCI) system allows one to communicate without any overt muscle movement. Electroencephalogram (EEG) is one of the most popular techniques to record brain activity. This paper proposes the multi-level voting method to classify brain activity depicting different types of imagery motor activity (Left Hand Movement, Right Hand Movement, Left Leg Movement and Right Leg Movement). The features were calculated using cross correlation. For multi-class classification, one verses rest approach was used. Four sets of classifiers were trained for each of the EEG channels and majority vote was calculated to get final class designation. The average classification accuracy of 86% was obtained.