Dual-LED imaging for finger liveliness detection and its evaluation with replicas

Area of Science:

• Biometrics
• Image Processing
• Forensic Science

Background:

• Unattended fingerprint identification systems require robust anti-spoofing measures.
• Detecting physiological changes, like finger color shifts, offers a promising liveness detection strategy.
• Previous methods using single-wavelength LEDs showed potential but required improvement in reliability.

Purpose of the Study:

• To investigate a dual-LED imaging system for enhanced liveness detection in fingerprint identification.
• To improve the reliability of detecting replicas by analyzing spectral color changes in live fingers.
• To establish criteria for differentiating between live finger data and replica data.

Main Methods:

• Development and implementation of a dual-LED imaging system using wavelengths at 530 nm and 630 nm.
• Recording 153 input trials involving nine types of fingerprint replicas and live fingers from 42 participants.
• Analysis of spectral data to identify characteristic changes indicative of a live finger.

Main Results:

• The dual-LED system successfully captured spectral data reflecting color changes in live fingers.
• Defined criteria were established to effectively separate data from live fingers and replicas.
• High reliability in distinguishing between genuine and fake fingerprint inputs was demonstrated.

Conclusions:

• A dual-LED imaging system offers a reliable method for liveness detection in fingerprint systems.
• Analyzing spectral color changes provides a robust signature to reject replica inputs.
• This approach significantly enhances the security of unattended biometric systems.