Logo des Repositoriums

Auflistung nach Autor:in "Campisi,Patrizio"

1 - 2 von 2
  • Textdokument
    Resting-state EEG: A Study on its non-Stationarity for Biometric Applicaions
    (BIOSIG 2017, 2017) Hine, Gabriel Emile; Maiorana, Emanuele; Campisi,Patrizio
    In the last years, several papers on EEG-based biometric recognition systems have been published. Specifically, most of the proposed contributions focus on brain signals recorded in resting state conditions, with either closed or open eyes. A common assumption is that the acquired signals are quasi-stationarity. In this paper, we investigate such property in terms of discriminative capability, and we analyze whether or not it holds throughout the entire duration of data collected over long periods. An extensive set of experimental tests, conducted over a database comprising signals collected from 50 subjects in three distinct acquisition sessions, shows that the most distinctive information of the brain signals is temporally located at the beginning of each recording.
  • Textdokument
    Steady-State Visual Evoked Potentials for EEG-Based Biometric Identification
    (BIOSIG 2017, 2017) Piciucco,Emanuela; Maiorana,Emanuele; Falzon,Owen; Camilleri,Kenneth P.; Campisi,Patrizio
    In this paper we propose a biometric recognition system based on steady-state visual evoked potentials (SSVEPs), exploiting brain signals elicited by repetitive stimuli having a constant frequency as identifiers. EEG responses to SSVEP stimuli flickering at different frequencies are recorded, and both mel-frequency cepstral coefficients (MFCCs) and autoregressive (AR) reflection coefficients are used as discriminative features of the enrolled users. An analysis of the permanence across time of the brain response to SSVEP stimuli is also performed, by exploiting EEG data acquired in sessions disjoint in time. The employed database is composed by EEG recordings taken from 25 healthy subjects during two different sessions with 15 day average distance between them. The results show that good recognition performance and a high level of permanence can be reached exploiting the proposed method.