A Study on ECG-based Biometrics using Open Source Hardware

Abstract

Biometric technology uses unique physical features as bioinformation. Electrocardiogram(ECG) is
a signal recording potential changes related to heartbeats in the form of voltage, and each individual
has unique ECG signals in terms of heart location and size thereof. Therefore, they are used to diagnose
diseases and examine personal identity. The ECG device can be made compact in comparison
with other biosignal measurement instruments and the signals measured in locations easily accessible.
This paper aims to implement a personal identification system based on ECG signals. Noise
removal and signal split is conducted as a pre-processing process by using the Band Pass Filter(BPF),
Median Filter(MF) and R-Peak. Features are extracted by using the morphological method from the
time domain. Euclidean Distance(ED) is used to analyze the verification data and registration data. In
the experiment result, For 18 verification data with the lowest EER 1.67%, driving was 113.4 seconds.