Multimodal Biometric Authentication through Peripheral Finger-Vein Patterns and Fingerprints
While multimodal biometrics have improved the security and reliability of biometric identity verification, current research efforts suffer from inaccuracy, high costs, and vulnerability to physical damage, making multimodal authentication difficult to implement in low-cost, real-world applications. To meet today’s biometric security needs, this project focuses on developing a bimodal biometric system based on inexpensive peripheral vein biometrics and fingerprints.
The system consists of a low-cost Near-Infrared camera, a low-power NIR LED array, and a fingerprint sensor. Through the Raspberry Pi Zero-W, the fingerprint sensor determines the fingerprint biometric template. The LEDs emit ~890 nanometer light, which is absorbed by the finger veins’ deoxygenated blood and detected by the camera. Then, a computer vision algorithm captures a raw image of the finger vein pattern, calculates the finger’s ROI using centroid analysis, performs Contrast-Limited-Adaptive-Histogram-Equalization to increase the veins’ contrast, and reduces image noise through a Bilateral Gaussian Filter. Finally, a binary threshold extracts the unique vein structure, which is then registered as a biometric template. MATLAB’s finger-vein normalized cross correlation and fingerprint minutiae matching algorithms compute a matching score [0-1] between the input biometrics and registered templates.
This process was repeated with all registered templates to test the system’s ability to use authentication input to identify individuals. A high-performance Support-Vector-Machine model accurately authenticates the user based on the biometric-match features.
The system captured, processed, and stored biometrics from 50 fingers, using templates to authenticate individuals from 1000 biometric samples. Future work includes developing a high-security SVM model from a larger test database.