Face feature ciphertext authentication system based on fully homomorphic encryption

With the widespread adoption of face recognition systems, identity authentication systems based on face recognition must fully consider the security of user face features. To prevent face feature leakage, face features cannot be manipulated and stored in clear text. Aiming at this problem, a face-character ciphertext authentication system based on fully homomorphic encryption is designed. The system performs face feature matching in the encrypted domain. At the same time, the database stores the homomorphic ciphertext of the face feature template. The whole system Authentication can be done without decrypting the face feature template. The system implementation part relies on Microsoft′s open source simple encryption algorithm library, and further explores a batch and dimension reduction scheme to balance the accuracy and computational complexity of face matching. The test results based on face reference database (LFW, IJB-A, IJB-B, CASIA) show that face-based ciphertext authentication based on fully homomorphic encryption is feasible and can basically meet the needs of real application scenarios.