Novel fiducial and non‐fiducial approaches to electrocardiogram‐based biometric systems. Issue 2 (1st June 2013)

• Record Type: Journal Article
• Authors: Pereira Coutinho, David; Silva, Hugo; Gamboa, Hugo; Fred, Ana; Figueiredo, Mário
• Is Part Of: IET biometrics. Volume 2:Issue 2(2013)
• Page Start: 64
• Page End: 75
• Publication Date: 2013-06-01
• Subjects: authorisation -- bioelectric phenomena -- biometrics (access control) -- data acquisition -- electrocardiography -- feature extraction -- performance evaluation -- real-time systems -- signal detection; nonfiducial approach -- noninvasive technique -- cardiac electrophysiological assessment -- functional diagnostics -- functional evaluation -- electrophysiological sensing -- robust signal acquisition devices -- ambulatory conditions -- ECG-based real-time biometric system -- human identification -- human authentication -- data acquisition -- signal processing -- feature extraction -- classification methodologies -- feature analysis methodologies -- performance evaluation -- clinical-grade ECG recording -- healthy control individuals -- computer activities -- computer usage scenario; Biometric identification -- Periodicals; 570.15195
• DOI: 10.1049/iet-bmt.2012.0055
• Languages: English
• ISSNs: 2047-4938

Abstract

Abstract : The electrocardiogram (ECG) is a non‐invasive and widely used technique for cardiac electrophysiological assessment. Although the ECG has traditionally only been used for functional diagnostic and evaluation, several advances in electrophysiological sensing have made available robust signal acquisition devices, particularly suited for ambulatory conditions, widening its range of applications. In particular, recent work has shown the potential of the ECG as a biometric trait, both for human identification and authentication. This study sets the ground for an ECG‐based real‐time biometric system. The authors describe an experimental setup and the evaluation of new fiducial and non‐fiducial approaches, including data acquisition, signal processing, feature extraction and analysis and classification methodologies, showing the applicability of the ECG as a real‐time biometric. Performance evaluation was done in clinical‐grade ECG recording from 51 healthy control individuals (of a publicly available benchmark dataset) as well as on data collected from 26 healthy volunteers performing computer activities without any posture or motion limitations, thus simulating a regular computer usage scenario.