EEG Biometrics: On the Use of Occipital Cortex Based Features from Visual Evoked Potentials
Keywords: biometrics, eeg, datasets, performance evaluation
AbstractThe potential of using Electro-Encephalo-Gram (EEG) data as a biometric identifier is studied. This is the first study that assesses looming stimuli for the creation of biometrically useful Visual Evoked Potentials (VEP), i.e. EEG responses due to visual stimuli. A novel method for the detection of VEP responses with minimal expert interaction is introduced. The EEG data, segmented based on the VEP, are used to create a reliable feature vector. In contrast to previous studies, we provide a publicly available evaluation dataset based on infants which is therefore not biased due to unhealthy individuals. Only data from the occipital cortex are used (i.e. about 3 of the many possible electrode positions in the scalp), making the potential EEG biometric capture devices relatively simpler.