ECG based human identification using Second Order Difference Plots.

Area of Science:

• Biometrics
• Signal Processing
• Cardiology

Background:

• Electrocardiogram (ECG) is a vital biometric signal with extensive research in signal analysis.
• ECG analysis is crucial for diagnosing cardiac diseases, demonstrating high classification performance.
• Continued advancements in signal analysis methods drive ongoing research in ECG applications.

Purpose of the Study:

• To explore the utilization of ECG signals for robust human identification.
• To propose and evaluate a novel quantification approach for ECG-based biometrics.
• To compare the proposed method with existing feature extraction techniques on Second Order Difference Plot (SODP).

Main Methods:

• Developed Logarithmic Grid Analysis, a new quantification method for SODP.
• Divided SODP regions into logarithmically increasing distances to analyze data point distribution.
• Extracted features from short-term (5-s) ECG signals across three distinct databases.
• Utilized k-Nearest Neighbor algorithm with 10-fold cross-validation for classification.

Main Results:

• Achieved high accuracy rates of 91.96%, 99.86%, and 95.12% for ECG-based human identification.
• Demonstrated the effectiveness of Logarithmic Grid Analysis in extracting detailed features from SODP.
• Validated the proposed method across multiple ECG databases.

Conclusions:

• The high density of data points at the center of SODP necessitates detailed analysis of central regions.
• Logarithmic Grid Analysis effectively captures detailed features by analyzing data point distribution in logarithmically scaled sub-regions.
• The method offers a compact feature set for accurate human identification using ECG.