Robust Identification and Segmentation of the Outer Skin Layers in Volumetric Fingerprint Data

Area of Science:

• Biometrics
• Medical Imaging
• Signal Processing

Background:

• Traditional fingerprint biometrics face challenges in acquisition (e.g., infants) and presentation attack detection (PAD).
• Existing systems are vulnerable to spoofing, limiting use in unsupervised authentication scenarios like border control.
• Optical coherence tomography (OCT) offers potential for improved fingerprint imaging.

Purpose of the Study:

• To develop a digital signal processing chain for extracting dual fingerprints from OCT fingertip scans.
• To segment both epidermal ('outer') and dermal ('inner') fingerprints from a single OCT scan.
• To convert these 3D fingerprints into a 2D grayscale format for minutiae extraction.

Main Methods:

• A novel digital signal processing pipeline was designed to segment outer and inner fingerprints from OCT data.
• 3D OCT fingerprint data was converted to 2D grayscale images compatible with existing minutiae extraction algorithms.
• The method was validated on two time-domain OCT scanners and processed large datasets efficiently using GPGPU computing.

Main Results:

• Dual fingerprints (outer and inner) were successfully extracted from OCT scans.
• Both OCT fingerprint types demonstrated backward compatibility with conventional 2D fingerprint recognition.
• The inner fingerprint, originating from the dermal layer, showed greater resilience to damage and enhanced reliability.

Conclusions:

• OCT-based dual fingerprinting presents a robust alternative to conventional biometrics.
• The inner fingerprint offers improved security and reliability, particularly for challenging acquisition conditions.
• This technology has significant implications for secure, unsupervised biometric authentication systems.