An in-depth study of ELEKTRA as a biometric system
In biometrics, more than 20 years evince that the EKG or ECG is a feasible method to perform user identification. In previous work, we presented ELEKTRA: an identification method using a heatmap of a set of aligned peaks of users’ EKG together with a simple CNN to perform user identification. With this work, we intend to show ELEKTRA’s adaptability and feasibility by testing our work over four different databases: the NSRDB, the MIT-BIHDB, the PTBDB and the GUDB. With these databases, user identification is performed over healthy users, over users with different cardiovascular diseases and even over users performing cardiovascular activities. We have achieved promising results with all databases regarding accuracy and error rates. In addition, we also conclude, thanks to the GUDB, that performing user identification over EKGs with users with higher heart rates is more complex.