Individual Biometric Identification Using Multi-Cycle Electrocardiographic Waveform Patterns

Area of Science:

• Biometrics
• Cardiology
• Signal Processing

Background:

• Electrocardiogram (ECG) waveforms reflect unique individual heart electrical properties.
• ECG patterns offer potential for biometric recognition applications.

Purpose of the Study:

• To present a new method for individual biometric identification using the entire ECG waveform pattern.
• To evaluate the effectiveness of different pattern matching algorithms for ECG-based biometrics.

Main Methods:

• Multi-cycle ECG signals were acquired using a 3-wire ECG circuit with electrodes on wrists or fingers.
• Four pattern matching algorithms (cosine similarity, cross correlation, city block, Euclidean distance) were tested.
• k-fold cross-validation was employed to assess generalization on independent datasets.

Main Results:

• Wrist measurements achieved up to 93.3% accuracy with three heartbeats for 55 subjects.
• Finger measurements demonstrated superior performance, reaching 99.3% with two heartbeats and 100% with three heartbeats for 20 subjects.

Conclusions:

• The proposed ECG waveform pattern matching method shows high potential for reliable individual biometric identification.
• ECG biometrics, particularly using finger measurements, offers a promising non-invasive identification approach.