Towards Fingerprint Spoofing Detection in the Terahertz Range.

Area of Science:

• Biometrics and Security
• Terahertz Spectroscopy
• Materials Science

Background:

• Fingerprint spoofing attacks pose a significant security risk to biometric systems.
• Non-intrusive detection methods are crucial for maintaining the integrity of identity verification.
• Understanding the stratum corneum's properties is key to differentiating real and fake fingerprints.

Purpose of the Study:

• To investigate the efficacy of terahertz time domain spectroscopy (TDS) for non-intrusive fingerprint spoofing detection.
• To characterize spoofing materials and compare their terahertz signatures with authentic fingerprints.
• To develop and evaluate novel detection algorithms based on terahertz data.

Main Methods:

• Utilized a terahertz time domain spectroscopy (TDS) setup in a reflection configuration.
• Characterized five spoofing materials and prepared thin/thick finger imitations.
• Determined the complex refractive index using TDS in transmission configuration and analyzed reflection signals in time and frequency domains.

Main Results:

• Terahertz signals from spoofing materials were distinct from those of authentic fingerprints.
• Measured signals were validated against theoretical calculations, with deconvolution explaining observed phenomena.
• Developed two detection methods (predefined features and deep learning) achieving high true detection rates of 87.9% and 98.8%.

Conclusions:

• Terahertz technology offers a viable solution for high-probability fingerprint spoofing detection.
• The distinct terahertz signatures of spoofing materials enable effective differentiation from real fingerprints.
• The proposed terahertz-based methods significantly enhance the security of biometric systems against spoofing attacks.