Cancelable ECG biometrics using GLRT and performance improvement using guided filter with irreversible guide signal

Area of Science:

• Biometrics and Security Engineering
• Signal Processing for Healthcare

Background:

• Biometric systems, including electrocardiogram (ECG) authentication, offer convenient individual verification but suffer from irrevocability if compromised.
• Cancelable biometrics aim to mitigate the risks associated with compromised biometric templates by enabling revocation and re-issuance.

Purpose of the Study:

• To propose a novel cancelable ECG biometrics system that overcomes the performance degradation often seen in such systems.
• To enhance the security and robustness of ECG-based authentication through advanced signal processing techniques.

Main Methods:

• Development of a cancelable ECG biometrics system utilizing a generalized likelihood ratio test (GLRT) detector derived from composite hypothesis testing in a randomly projected domain.
• Introduction of a guided filtering (GF) technique with an irreversible guide signal for ECG authentication templates to improve performance.
• Evaluation of the proposed method on the ECG-ID database, comparing it against conventional detectors with original and compressed ECG data.

Main Results:

• The proposed GLRT detector with 10% compressed ECG achieved 91.4% PD1, outperforming the Euclidean detector with the same compressed data (90.8% PD1).
• Guided filtering with the irreversible ECG template further boosted the GLRT detector's performance to 94.3% PD1, surpassing the conventional Euclidean detector with original ECG (93.9% PD1).
• The proposed cancelable ECG biometrics system demonstrated efficiency, re-usability, diversity, and non-invertibility, meeting key cancelable biometrics criteria.

Conclusions:

• The proposed cancelable ECG biometrics system, integrating GLRT and guided filtering, effectively enhances security and performance.
• The method provides a robust solution to the irrevocability issue in biometrics, offering a practical and secure authentication alternative.
• The system meets essential cancelable biometrics requirements, paving the way for more secure and adaptable biometric authentication solutions.