Fusion of ECG sources for human identification

A study about the applicability of electrocardiogram (ECG) signals in human identification systems is presented in this paper. There is strong evidence that ECG signals embed highly discriminative information in a population. In this paper, multiple levels of fusion are described to combine information gained from the standard 12 lead ECG system. Most of the current approaches make use of only one lead electrocardiogram to extract fiducial based features. However, in pattern recognition problems it is believed that the larger the amount of information, the higher the probability of recognizing a subject successfully. The primary goal of the current work is to demonstrate that all ECG leads are suitable for identification and to suggest feature and decision level fusion techniques of combining this data. When the proposed system is tested on a public dataset, considerably high recognition rates are achieved.