A Secure Biometric Template Protection Mechanism Against Similarity-based Attack
Yazhou Wang, Bing Li, Jiaxin Wu, Qianya Ma, Guozhu Liu, Yuqi Li · 2022 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS) · 2022
Biometric features have been widely applied in the field of identity authentication. Non-invertible biometric template protection methods (BTP) are popularly used to protect biometric traits in an authentication system. However, biometric features can be reconstructed from the transformed biometric templates in the database by similarity-based attack (SA), which leads to the privacy and security issues in the authentication system. To address these issues, we design a secure biometric template protection mechanism via a many-to-one mapping approach. In our scheme, on the one hand, the combination of random permutation and inner product is proposed to achieve a many-to-one mapping relationship between biometric feature and biometric template, which is used to resist SA. On the other hand, we propose a framework for evaluating information leakage between the original space and hashed space. We conducted the SA experiments to verify the effectiveness of our scheme when compared to related works. Experimental results show our scheme not only improve the security against similarity-based attack but also effectively preserve the accuracy performance. Besides, we also analyze non-linkability, revocability, non-invertibility, and accuracy preservation, which proves our scheme meets BTP standard properties.