ECG slope features for Biometric Authentication

Abstract

This paper presents a new feature extraction technique, based on the slope of Electrocardiograms (ECG), for biometric authentication. An algorithm has been developed based on the concept of amplitude comparison between a triangle wave and the ECG signal. The proposed algorithm performance has been evaluated by comparing the Receiver Operating Characteristic (ROC) curve for various triangular wave settings. 1000 ECGs recorded from 100 students have been used in this work. Different techniques in producing triangular waves to extract the ECG features have been presented in this paper. It is shows in this paper that the authentication performance increases 50% when the features are extracted separately from the peaks of P to R and from the peaks of R to T.