ECG biometric identification: A compression based approach.

Area of Science:

• Biometrics
• Signal Processing
• Information Theory

Background:

• Current electrocardiogram (ECG) biometrics face challenges in accuracy and robustness.
• ECG signals offer unique, liveliness-dependent characteristics for identification.
• Traditional methods are susceptible to interferences due to heartbeat segmentation.

Purpose of the Study:

• To develop an innovative and robust solution for person identification and authentication using ECG signals.
• To propose a novel similarity measure based on information-theoretic data models and Kolmogorov complexity approximation.
• To overcome limitations of traditional ECG biometrics, such as reliance on heartbeat segmentation.

Main Methods:

• Utilized information-theoretic data models for data compression.
• Employed similarity metrics related to Kolmogorov complexity approximation for ECG segment comparison.
• Implemented a leave-one-out cross-validation strategy with a 1-Nearest Neighbor (1NN) classifier.

Main Results:

• The proposed measure effectively clusters ECG data into identical records, same participant, and different participant groups.
• Achieved high accuracy (99%) in identifying participants within a database using the novel distance measure.
• Demonstrated robustness by avoiding traditional heartbeat segmentation.

Conclusions:

• The developed information-theoretic approach provides a robust and accurate method for ECG-based biometrics.
• The technique offers a significant improvement over existing ECG identification methods.
• This approach holds promise for secure and reliable person authentication.