ECG analysis for human recognition using non‐fiducial methods

Abstract

The electrocardiogram (ECG) has emerged as a new biometric for human recognition due to its robustness against fraudulent attacks. This article presents a novel method of ECG biometric for human recognition using autocorrelation (AC) followed by one of the three transformation techniques, i.e. discrete cosine transform (DCT), discrete Fourier transform (DFT), and Walsh–Hadamard transform (WHT). The effectiveness of these transformations is evaluated on the dimensionality reduction techniques i.e. principal component analysis and linear discriminant analysis (LDA). Thus, the systems prepared by different combinations of transformations and dimensionality reduction techniques are evaluated on publically available databases of Physionet. The authentication and identification accuracies achieved by these systems are found the best on DFT and LDA combination. The authentication performance is reported to 99.98% (99.83%), whereas the average rank classification accuracy is reported to 100% (97%) on QT database (MIT-BIH arrhythmia database) of Physionet.