DescriptionThe use of brain activity as physiological basis for biometric systems has been receiving increased attention in recent years, as it can offer attractive properties such as robustness against spoo fing attacks and liveness detection. An imaging modality that can be used to acquire brain activity is electroencephalography (EEG). A major challenge in EEG-based biometric systems, however, is to identify reliable signatures of individuality from the acquired EEG data, that are also invariant against time. Motivated by prior neuroscience studies reporting large inter-individual variability in the spectral fi eld powers of EEG recordings, in this Thesis, we propose a method for extracting EEG-based signatures of individuality, and investigate their invariability against time. Features are extracted based on the spatial distribution of the spectral power of EEG data, corresponding to 2-second eyes-closed resting-state (ECRS) recording, across di fferent bands. ECRS EEG data in 4 healthy volunteers are recorded in two di fferent sessions with an interval of at least one week between sessions, using a 128-channel system. To investigate the invariability of the proposed features against time, identi cation accuracy is examined for two scenarios: 1) the training and testing datasets are chosen from the same recording session, and 2) the training dataset is chosen from one session, and the testing dataset is chosen from another session. For the first scenario, an identi cation accuracy of 99.1% is achieved when the proposed features are extracted from the beta2 frequency band. For the second scenario, an identification accuracy of 92.3% is achieved when the proposed features from both alpha and beta frequency bands are used. To improve collectability, recording channels that carry the most discriminatory information across individuals are identi fied using principal component analysis (PCA). 48 channels, mostly covering the occipital region are chosen. Identifi cation accuracy similar to what was obtained for the case of 128 channels, is achieved for both testing scenarios. The results of this work suggest that features based on the spatial distribution of the spectral power of the short-time ECRS recordings can have great potentials in EEG-based biometric identi fication systems.