The Invisible Threat: Evaluating the Vulnerability of Cross-Spectral Face Recognition to Presentation Attacks

2025-05-01

AI Summary

Near-infrared–visible (NIR-VIS) cross-spectral face recognition is widely assumed to be inherently robust against presentation attacks (PAs), yet its real-world vulnerability remains systematically unassessed. Method: This work conducts the first ISO/IEC 30107-3–compliant empirical evaluation of NIR-VIS systems against diverse PAs—including printed photos, screen replay, and 3D masks—using state-of-the-art deep cross-spectral models (e.g., AENet, CMFA) and a newly collected, realistic NIR-VIS paired dataset. Contribution/Results: High-fidelity screen replay and 3D mask attacks degrade genuine acceptance rates by over 40% in commercial-grade systems, revealing critical security blind spots beneath claimed robustness. The study establishes the first standardized benchmark framework for PA detection in NIR-VIS cross-spectral biometrics, providing both methodological guidance and empirical evidence for rigorous security assessment of cross-spectral recognition systems.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Application Domains: Security

Application Category

• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Cross-spectral face recognition systems are designed to enhance the performance of facial recognition systems by enabling cross-modal matching under challenging operational conditions. A particularly relevant application is the matching of near-infrared (NIR) images to visible-spectrum (VIS) images, enabling the verification of individuals by comparing NIR facial captures acquired with VIS reference images. The use of NIR imaging offers several advantages, including greater robustness to illumination variations, better visibility through glasses and glare, and greater resistance to presentation attacks. Despite these claimed benefits, the robustness of NIR-based systems against presentation attacks has not been systematically studied in the literature. In this work, we conduct a comprehensive evaluation into the vulnerability of NIR-VIS cross-spectral face recognition systems to presentation attacks. Our empirical findings indicate that, although these systems exhibit a certain degree of reliability, they remain vulnerable to specific attacks, emphasizing the need for further research in this area.

Problem

• Evaluating vulnerability of cross-spectral face recognition to attacks
• Assessing NIR-VIS system robustness against presentation attacks
• Identifying reliability gaps in NIR-based facial recognition

Innovation

• Cross-spectral NIR-VIS face recognition for robust matching
• Evaluating vulnerability to presentation attacks systematically
• NIR imaging enhances resistance to illumination variations