Online Biometric Authentication Using Subject-Specific Band Power features of EEG

Abstract

Biometric recognition of persons based on unique features extracted from brain signals is an emerging area of research nowadays, on account of the subject-specificity of human neural activity. This paper proposes an online Electroencephalogram (EEG) based biometric authentication system using band power features extracted from alpha, beta and gamma bands, when the subject is in relaxed rest state with eyes open or closed. The most distinct band features are chosen specifically for each subject which are then used to generate subject-specific template during enrollment. During online authentication, recorded test EEG pattern is matched with the respective template stored in the database and degree of matching in terms of its correlation coefficient predicts the genuineness of the claimant. A number of client and imposter authentication tests have been conducted in online framework among 6 subjects using the proposed system, and achieves an average recognition rate of 88.33% using 14 EEG channels. Experimental analysis shows the subject-specificity of distinct bands and features, and highlights the utility of subject-specific band power features in EEG-based biometric systems.