| Title |
A Study on Baseline Zero-point Calibration Technique of ECGfor User Authentication |
| Authors |
최규호(Gyu Ho Choi) ; 반성범(Sung Bum Pan) |
| DOI |
https://doi.org/10.5573/ieie.2020.57.4.63 |
| Keywords |
ECG; Baseline zero-point calibration; Regression analysis; User authentication |
| Abstract |
Recently, user authentication technology using ECG, which is stronger in forgery and alteration than fingerprint and face authentication, has been actively studied. The ECG signal is an electrical signal that contains several noises. In particular, as the baseline fluctuation noise generated from the subject's breathing becomes more severe, user authentication performance is degraded due to morphological characterization due to nonlinear noise. Although frequency filters are used to remove baseline fluctuation noise, there is a problem of distorting the morphological characteristics of the ECG. In this paper, baseline variation is estimated by first-order regression analysis using median T-P wave and P-R wave in ECG, and the estimated baseline is adjusted to zero. Experimental results showed that the average similarity was improved by up to 4.6% in 10 cycles of ECG after baseline zeroing. The maximum EER performance of user authentication before baseline zeroing was 1.82%, but the maximum performance after baseline zeroing was improved to 1.11%. |