Template recovery attack on encrypted face recognition systems with unprotected decision using synthetic faces

IntroductionHomomorphic encryption (HE) enables privacy-preserving face recognition by allowing encrypted facial embeddings to be compared without decryption. While efficient, these systems often reveal comparison scores in plaintext, introducing a security risk. Revealing these scores can potential...

Full description

Saved in:
Bibliographic Details
Main Authors: Amina Bassit, Florian Hahn, Zohra Rezgui, Hatef Otroshi Shahreza, Raymond Veldhuis, Andreas Peter
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-05-01
Series:Frontiers in Imaging
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fimag.2025.1476377/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850189773743849472
author Amina Bassit
Amina Bassit
Florian Hahn
Zohra Rezgui
Hatef Otroshi Shahreza
Hatef Otroshi Shahreza
Raymond Veldhuis
Raymond Veldhuis
Andreas Peter
Andreas Peter
author_facet Amina Bassit
Amina Bassit
Florian Hahn
Zohra Rezgui
Hatef Otroshi Shahreza
Hatef Otroshi Shahreza
Raymond Veldhuis
Raymond Veldhuis
Andreas Peter
Andreas Peter
author_sort Amina Bassit
collection DOAJ
description IntroductionHomomorphic encryption (HE) enables privacy-preserving face recognition by allowing encrypted facial embeddings to be compared without decryption. While efficient, these systems often reveal comparison scores in plaintext, introducing a security risk. Revealing these scores can potentially allow adversaries to reconstruct sensitive facial embeddings and infer demographic attributes, thus compromising user privacy.MethodsThis work proposes a training-less face template recovery attack leveraging the Lagrange multiplier optimization method. The attack requires only a small set of randomly generated synthetic facial images and their associated comparison scores with a target template. The method assumes attackers use spoofed synthetic faces and lack direct access to the face recognition system, aligning with real-world threat models.ResultsExperimental evaluation demonstrates the feasibility and effectiveness of the proposed attack. It shows that between 50 and 192 comparison scores and synthetic images are sufficient to recover the target face template with 100% success under strict system thresholds. The recovered templates closely resemble the original and retain identifiable soft biometric traits.DiscussionThe findings reveal a critical vulnerability in face recognition systems employing inner product similarity measures under homomorphic encryption. Even without system access or training data, attackers can exploit leaked comparison scores to compromise facial privacy. The study underscores the need to reassess how score leakage is handled in encrypted recognition systems and explore stronger protection mechanisms against template reconstruction.
format Article
id doaj-art-c1e5dacc0f3c4a95acda8990430d3e26
institution OA Journals
issn 2813-3315
language English
publishDate 2025-05-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Imaging
spelling doaj-art-c1e5dacc0f3c4a95acda8990430d3e262025-08-20T02:15:32ZengFrontiers Media S.A.Frontiers in Imaging2813-33152025-05-01410.3389/fimag.2025.14763771476377Template recovery attack on encrypted face recognition systems with unprotected decision using synthetic facesAmina Bassit0Amina Bassit1Florian Hahn2Zohra Rezgui3Hatef Otroshi Shahreza4Hatef Otroshi Shahreza5Raymond Veldhuis6Raymond Veldhuis7Andreas Peter8Andreas Peter9Department of Computer Science and Engineering, Michigan State University, East Lansing, MI, United StatesData Management & Biometrics and Semantics, Cybersecurity & Services, University of Twente, Enschede, NetherlandsData Management & Biometrics and Semantics, Cybersecurity & Services, University of Twente, Enschede, NetherlandsData Management & Biometrics and Semantics, Cybersecurity & Services, University of Twente, Enschede, NetherlandsBiometrics Security & Privacy Group, Idiap Research Institute, Martigny, SwitzerlandSchool of Engineering, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, SwitzerlandData Management & Biometrics and Semantics, Cybersecurity & Services, University of Twente, Enschede, NetherlandsDepartment of Information Security and Communication Technology, Norwegian University of Science and Technology, Gjøvik, NorwayData Management & Biometrics and Semantics, Cybersecurity & Services, University of Twente, Enschede, NetherlandsSafety-Security-Interaction Group, Carl von Ossietzky Universität Oldenburg, Oldenburg, GermanyIntroductionHomomorphic encryption (HE) enables privacy-preserving face recognition by allowing encrypted facial embeddings to be compared without decryption. While efficient, these systems often reveal comparison scores in plaintext, introducing a security risk. Revealing these scores can potentially allow adversaries to reconstruct sensitive facial embeddings and infer demographic attributes, thus compromising user privacy.MethodsThis work proposes a training-less face template recovery attack leveraging the Lagrange multiplier optimization method. The attack requires only a small set of randomly generated synthetic facial images and their associated comparison scores with a target template. The method assumes attackers use spoofed synthetic faces and lack direct access to the face recognition system, aligning with real-world threat models.ResultsExperimental evaluation demonstrates the feasibility and effectiveness of the proposed attack. It shows that between 50 and 192 comparison scores and synthetic images are sufficient to recover the target face template with 100% success under strict system thresholds. The recovered templates closely resemble the original and retain identifiable soft biometric traits.DiscussionThe findings reveal a critical vulnerability in face recognition systems employing inner product similarity measures under homomorphic encryption. Even without system access or training data, attackers can exploit leaked comparison scores to compromise facial privacy. The study underscores the need to reassess how score leakage is handled in encrypted recognition systems and explore stronger protection mechanisms against template reconstruction.https://www.frontiersin.org/articles/10.3389/fimag.2025.1476377/fullhomomorphic encryptiontemplate recoverybiometric template protectioninner product-based scorebiometric recognitionsynthetic facial images
spellingShingle Amina Bassit
Amina Bassit
Florian Hahn
Zohra Rezgui
Hatef Otroshi Shahreza
Hatef Otroshi Shahreza
Raymond Veldhuis
Raymond Veldhuis
Andreas Peter
Andreas Peter
Template recovery attack on encrypted face recognition systems with unprotected decision using synthetic faces
Frontiers in Imaging
homomorphic encryption
template recovery
biometric template protection
inner product-based score
biometric recognition
synthetic facial images
title Template recovery attack on encrypted face recognition systems with unprotected decision using synthetic faces
title_full Template recovery attack on encrypted face recognition systems with unprotected decision using synthetic faces
title_fullStr Template recovery attack on encrypted face recognition systems with unprotected decision using synthetic faces
title_full_unstemmed Template recovery attack on encrypted face recognition systems with unprotected decision using synthetic faces
title_short Template recovery attack on encrypted face recognition systems with unprotected decision using synthetic faces
title_sort template recovery attack on encrypted face recognition systems with unprotected decision using synthetic faces
topic homomorphic encryption
template recovery
biometric template protection
inner product-based score
biometric recognition
synthetic facial images
url https://www.frontiersin.org/articles/10.3389/fimag.2025.1476377/full
work_keys_str_mv AT aminabassit templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT aminabassit templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT florianhahn templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT zohrarezgui templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT hatefotroshishahreza templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT hatefotroshishahreza templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT raymondveldhuis templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT raymondveldhuis templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT andreaspeter templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces
AT andreaspeter templaterecoveryattackonencryptedfacerecognitionsystemswithunprotecteddecisionusingsyntheticfaces