Secure and reversible face anonymization with diffusion models

2025-10-01

AI Summary

Existing diffusion-based face anonymization methods struggle to simultaneously achieve high image fidelity and secure reversibility: they either lack key-based control—posing risks of identity leakage upon inversion—or sacrifice visual quality. This paper proposes the first latent-space-oriented, key-driven reversible diffusion framework. It embeds cryptographic keys into both the forward and reverse diffusion processes and incorporates facial mask constraints to preserve identity-agnostic features, thereby improving the privacy–quality trade-off. The method enables deterministic, key-dependent anonymization and authorized reconstruction directly in the latent space. Experiments demonstrate significantly reduced visual distortion compared to state-of-the-art methods; critically, original identities can be accurately recovered *only* by users possessing the correct decryption key. Thus, the approach achieves strong privacy guarantees without compromising practical utility.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Privacy
• Philosophy and Ethics of AI: Privacy & Security

Application Category

• Security and Privacy: Data transparency and provenance
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems

Abstract

Face images processed by computer vision algorithms contain sensitive personal information that malicious actors can capture without consent. These privacy and security risks highlight the need for effective face anonymization methods. Current methods struggle to propose a good trade-off between a secure scheme with high-quality image generation and reversibility for later person authentication. Diffusion-based approaches produce high-quality anonymized images but lack the secret key mechanism to ensure that only authorized parties can reverse the process. In this paper, we introduce, to our knowledge, the first secure, high-quality reversible anonymization method based on a diffusion model. We propose to combine the secret key with the latent faces representation of the diffusion model. To preserve identity-irrelevant features, generation is constrained by a facial mask, maintaining high-quality images. By using a deterministic forward and backward diffusion process, our approach enforces that the original face can be recovered with the correct secret key. We also show that the proposed method produces anonymized faces that are less visually similar to the original faces, compared to other previous work.

Problem

• Securing face anonymization with reversible key-based diffusion models
• Balancing privacy protection and image quality in face processing
• Enabling controlled reversibility for authorized person authentication

Innovation

• Combines secret key with diffusion model latent representation
• Uses facial mask to preserve identity-irrelevant image features
• Employs deterministic diffusion process for reversible face recovery