Generic Biometry Algorithm Based on Signal Morphology Information: Application in the Electrocardiogram Signal
This work presents the development, test, and implementation of a new biometric identification procedure based on electrocardiogram (ECG) signal morphology. ECG data were collected from 63 subjects during two data-recording sessions separated by six months (Time Instance 1, T1, and Time Instance 2, T2). Two tests were performed aiming at subject identification, using a distance-based method with the heartbeat patterns. In both tests, the enrollment template was composed by the averaging of all the T1 waves for each subject. Two testing datasets were created with five meanwaves per subject. While in the first test the meanwaves were composed with different T1 waves, in the second test T2 waves were used. The T2 waves belonged to the same subjects but were acquired in different time instances, simulating a real biometric identification problem. The classification was performed through the implementation of a kNN classifier, using the meanwave’s Euclidean distances as the features for subject identification. The accuracy achieved was 95.2 % for the first test and 90.5 % for the second. These results were achieved with the optimization of some crucial parameters. In this work we determine the influence of those parameters, such as, the removal of signal outliers and the number of waves that compose the test meanwaves, in the overall algorithm performance. In a real time identification problem, this last parameter is related with the length of ECG signal needed to perform an accurate decision. Concerning the study here depicted, we conclude that a distance-based method using the subject’s ECG signal morphology is a valid parameter for classification in biometric applications.
KeywordsBiometry Classification Electrocardiography Meanwave Signal processing
The authors would like to thank the Escola Superior de Saúde-Cruz Vermelha Portuguesa (ESSCVP) for the data collections infrastructures and subjects providence.
- 2.Silva, H., Gamboa, H., Fred, A.: Applicability of lead v2 ecg measurements in biometrics. In: Proceedings of Med-e-Tel (2007)Google Scholar
- 3.Coutinho, D. P., Fred, A. L. N., Figueiredo, M. A. T.: Personal identification and authentication based on one-lead ecg using ziv-merhav cross parsing. In: 10th International Workshop on Pattern Recognition in Information Systems (2010)Google Scholar
- 4.Li, M., Narayanan, S.: Robust ecg biometrics by fusing temporal and cepstral information. In: 20th International Conference on Pattern Recognition (2010)Google Scholar
- 5.Lourenco, A., Silva, H., Fred, A.: Unveiling the biometric potential of finger-based ecg signals. In: Computational Intelligence and Neuroscience (2011)Google Scholar
- 6.Plataniotis, K.N., Hatzinakos, D., Lee, J.K.M.: Ecg biometric recognition without fiducial detection. In: Biometric Consortium Conference, Biometrics Symposium (2006)Google Scholar
- 7.Nunes, N., Araujo, T., Gamboa, H.: Time series clustering algorithm for two-modes cyclic biosignals. In: Fred, A., Filipe, J., Gamboa, H. (eds.) BIOSTEC 2011, CCIS 273, pp. 233–245. Springer, Heidelberg (2012)Google Scholar
- 8.Orange. http://orange.biolab.si/ (2012)
- 9.Chan, A.D.C., Hamdy, M.M., Badre, A., Badee, V.: Wavelet distance measure for person identification using electrocardiograms in IEEE Transactions on Instrumentation and Measurement (2008)Google Scholar