Reliable features for an ECG-based biometric system

Verification of subjects using their unique physiological features has recently attracted much attention to develop secure biometric systems. One of the most reliable physiological features is electrocardiogram (ECG) waveform, which is the electrical reflection of the heart activity, and has a unique characteristic for each individual. In this paper, autoregressive (AR) coefficients along with mean of power spectral density (PSD) were used as reliable ECG features to enhance the performance of an ECG-based biometric system. To assess the effectiveness of the proposed combination, other features including autoregressive (AR) coefficients, Higuchi dimension, Lyapunov exponent, and approximation entropy (ApEn) were extracted from ECG Multi-layer-perceptron (MLP), probabilistic neural networks, and k-nearest neighbor (KNN) classifiers were used to classify the extracted features. In addition, simple combination of the features was considered for further improvement in verification rate. The achieved results (100% accuracy) showed the effectiveness of the combined features in terms of accuracy and robustness compared to the results produced by the former traditional methods.