Medical biometrics: the perils of ignoring time dependency

Authors:
Foteini Agrafioti;Francis M. Bui;Dimitrios Hatzinakos
Affiliations:
Edward S. Rogers Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON, Canada;Edward S. Rogers Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON, Canada;Edward S. Rogers Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON, Canada
Venue:
BTAS'09 Proceedings of the 3rd IEEE international conference on Biometrics: Theory, applications and systems
Year:
2009

Citing 6
Cited 0

Online stress detection using psychophysiological signals for implicit human-robot cooperation

Robotica
Analysis of human electrocardiogram for biometric recognition

EURASIP Journal on Advances in Signal Processing
Unobtrusive multimodal biometric authentication: the HUMABIO project concept

EURASIP Journal on Advances in Signal Processing
On supporting anonymity in a BAN biometric framework

DSP'09 Proceedings of the 16th international conference on Digital Signal Processing
ECG to identify individuals

Pattern Recognition
A receiver-based variable-size-burst equalization strategy for spectrally efficient wireless communications

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

The electrocardiogram (ECG) is a medical signal that has lately drawn interest from the biometrics community, and has been shown to have significantly discriminative characteristics in a population. This paper brings to light the particular challenges of electrocardiogram recognition to advocate that time dependency is a controversial point. In contrast to traditional biometrics, ECG allows for continuous authentication and consequently expands the range of applications. However, time varying biometrics put on the line the recognition accuracy due to increased intra subject variability. This paper suggests a novel framework for bypassing this inadequacy. A template update methodology is proposed and demonstrated to boost the recognition performance over 2 hour recordings of 10 subjects.