ECG Biometrics: New Algorithm and Multimodal Biometric System

This thesis studies the challenges of applying the electrocardiogram signal (ECG) as a biometric. Unlike most biometric characteristics, ECG is time-dependent and naturally affected by the physical and psychological activity of the human body which is mainly neglected by the ECG biometric community. In this work we first presents the UofT ECG database (UofTDB) which offers a comprehensive analysis of the underlying interindividual variability under a number of conditions, such as body posture, physical activity, and time lapse. Then we propose a new biometric methodology and evaluate its performance using the UofTDB. The new method outperforms the state of the art ECG method AC/LDA and is robust to day-to-day ECG variations (3.12% EER). Finally we investigate the fusion of ECG and fingerprint and propose a fusion scheme to be robust to spoof attacks and achieves a very promising performance for real life applications (~0.08% EER).M.A.S