GFT-GCN: Privacy-Preserving 3D Face Mesh Recognition with Spectral Diffusion

2025-11-25

AI Summary

To address the vulnerability of 3D facial biometric templates to reconstruction attacks and the inherent trade-off between privacy preservation and recognition accuracy, this paper proposes a spectral diffusion-based template protection mechanism. At the client side, compact and discriminative spectral features are extracted via Graph Fourier Transform (GFT) and Graph Convolutional Networks (GCNs) from raw 3D face data; subsequently, an irreversible, renewable, and unlinkable protected template is generated through a spectral-domain diffusion process—ensuring that original 3D facial data never leaves the device. The method adopts a lightweight client–server architecture and achieves high recognition accuracy (>98.5%) on BU-3DFE and FaceScape benchmarks, while significantly enhancing robustness against template reconstruction attacks. Its core contribution lies in the first integration of spectral graph learning with diffusion modeling for biometric template protection, enabling simultaneous optimization of privacy, security, and recognition performance.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Privacy
• Natural Language Processing: Safety and Robustness

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

3D face recognition offers a robust biometric solution by capturing facial geometry, providing resilience to variations in illumination, pose changes, and presentation attacks. Its strong spoof resistance makes it suitable for high-security applications, but protecting stored biometric templates remains critical. We present GFT-GCN, a privacy-preserving 3D face recognition framework that combines spectral graph learning with diffusion-based template protection. Our approach integrates the Graph Fourier Transform (GFT) and Graph Convolutional Networks (GCN) to extract compact, discriminative spectral features from 3D face meshes. To secure these features, we introduce a spectral diffusion mechanism that produces irreversible, renewable, and unlinkable templates. A lightweight client-server architecture ensures that raw biometric data never leaves the client device. Experiments on the BU-3DFE and FaceScape datasets demonstrate high recognition accuracy and strong resistance to reconstruction attacks. Results show that GFT-GCN effectively balances privacy and performance, offering a practical solution for secure 3D face authentication.

Problem

• Protecting stored biometric templates in 3D face recognition systems
• Securing spectral features with irreversible and unlinkable templates
• Balancing privacy protection with high recognition accuracy performance

Innovation

• Graph Fourier Transform with Graph Convolutional Networks for feature extraction
• Spectral diffusion mechanism to create irreversible biometric templates
• Lightweight client-server architecture protecting raw biometric data