ECG Based Human Identification Using Random Forests
Résumé
Security concerns increase as the technology for falsification advances. There are strong evidences that a difficult to falsify biometric trait, the human heartbeat, can be used for identity recognition. Traditional solutions for biometric recognition from electrocardiogram (ECG) signals are limited in its power. They are based on temporal and amplitude distances between detected fiducial points. The current fiducial detection tools are inadequate for this application since the boundaries of waveforms are difficult to detect, locate and define. In this study, the ECG signals were used to identify a total of 120 individuals obtained from four ECG databases obtained from the Physionet database (MIT-BIH, ST-T, NSR, PTB) and an ECG database collected from 40 student volunteers from Paris Est University. Feature extraction from the ECG signals was performed by using Discrete Wavelet Transform (DWT). The Random Forest was then presented for the ECG signals identification. Preliminary experimental results indicate that the system is accurate and can achieve a low false negative rate, low false positive rate and a 100% subject recognition rate for healthy subjects with the reduced set of features.