The Proposal and It’s Evalution of Biometric Authentication Method by EEG Analysis Using Image Stimulation

摘要

In recent years, techniques of Brain Machine Interface (BMI) which conducts human communication and robot manipulation using human brain activity are widely researched. This is the result of a noninvasive electroencephalograph device that can measure Electroencephalogram (EEG) in real time. However, there is a present condition that the authentication method when BMI is not much researched. In our research, we propose a biometric authentication method of electroencephalogram using image stimulation. In this research, we propose a biometric authentication method of electroencephalogram using image stimulation. In this paper, we construct and then evaluate a system that performs biometric authentication using EEG at image stimulus. We perform feature extraction using cross-correlation coefficient, and SVM for classification / authentication. Moreover We considered the method for preprocessing (digital filter, artifact countermeasure, epoch), we verify more appropriate preprocessing method. We verified the proposed method. In our proposed system, EER: 2.0% was obtained when artifact countermeasure, digital filter (IIR filter), and epoch method were used. From the result of FAR and FRR, our system was suggested that accuracy is improved by taking artifact countermeasure.