Morphological synthesis of ECG signals for person authentication

In this paper, we propose a novel approach using the morphological properties of the R-R segments in an ECG signal for biometric authentication. The authenticity of a given R-R segment is decided by comparing it to a matching R-R segment morphologically synthesized from a model characterizing the identity to be authenticated. The morphological synthesis process uses a set of R-R segments (templates), recorded during enrolment at different heart-rates, and a time alignment algorithm. This ensures the authentication to be independent of the, usually variable, heart-rate. The optimum average equal-error-rate obtained in our experiments is 2%.