Analysis of Human Electrocardiogram for Biometric Recognition Using Analytic and AR Modeling Extracted Parameters

The electrocardiograph (ECG) contains cardiac features unique to each individual. By analyzing ECG, it should therefore be possible not only to detect the rate and consistency of heartbeats but to also extract other signal features in order to identify ECG records belonging to individual subjects. In this paper, a new approach for automatic analysis of single lead ECG for human recognition is proposed and evaluated. Eighteen temporal, amplitude, width and autoregressive (AR) model parameters are extracted from each ECG beat and classified in order to identify each individual. Proposed system uses pre-processing stage to decrease the effects of noise and other unwanted artifacts usually present in raw ECG data. Following pre-processing steps, ECG stream is partitioned into separate windows where each window includes single beat of ECG signal. Window estimation is based on the localization of the R peaks in the ECG stream that detected by Filter bank method for QRS complex detection. ECG features – temporal, amplitude and AR coefficients are then extracted and used as an input to K-nn and SVM classification algorithms in order to identify the individual subjects and beats. Signal pre-processing techniques, applied feature extraction methods and some intermediate and final classification results are presented in this paper.