EEG authentication system based on auto-regression coefficients

Researches on brain signals have indicated that electroencephalogram (EEG) signals of different persons under the same experimental environment have different identity characteristics. In order to better utilize EEG signals in identity authentication, a new analytical approach to study authentication system based on EEG signals is designed in this study. To increase the stability of authentication procedure, event related potentials (ERPs) are introduced in this study. Subjects' photos are used as one of stimulus source and subjects are divided into groups when their EEG signals are collected. To increase classification accuracy, auto-regression coefficients methods are used in this study to extract features.