On establishing evaluation standards for ECG biometrics

Electrocardiogram (ECG) biometrics are becoming increasingly popular. Numerous approaches to ECG processing have been proposed over the past years and the field has drawn significant attention from the biometrics community. However, less attention has been paid to developing a standard for ECG biometric testing for the evaluation of such algorithms. This paper proposes a set of standards for ECG signal recording and presents the UofT ECG Database (UofTDB) in order to evaluate the performance of various ECG biometric methods. Compared to existing databases, the UofTDB has three important characteristics namely, large population size (1012 individuals), varying body postures, physical exercise and acquisition over a long period. A promising equal error rate of under 5% is also reported.