System for secure image recognition
Disclosed embodiments include methods, apparatuses, and systems for secured image processing, image recognition, biometric recognition, and face recognition in untrusted environments. The disclosure includes a system for secure image recognition that comprises a secure biometric recognition system configured to work directly with encrypted signals, and the secure biometric recognition system comprises an input quantization system and a homomorphic encryption system configured for noninteractive biometric recognition.
1. A system for secure image recognition comprising a secure biometric recognition system configured to work directly with encrypted templates and encrypted query features, said secure biometric recognition system comprising an input order-preserving quantization system and a homomorphic encryption system configured for noninteractive biometric recognition in a single communication round using a hardware processor,
wherein:
a first party utilizes the input order-preserving quantization system to quantify a query feature vector and utilizes the homomorphic encryption system to encrypt the quantized query feature vector,
the first party sends the quantized and encrypted query feature vector to a second party that owns a database of encrypted templates but cannot decrypt them and cannot decrypt the received quantized and encrypted query feature vector,
the second party utilizes the homomorphic encryption system to calculate an encrypted biometric recognition result using the encrypted templates and the quantized and encrypted query feature vector,
the second party sends the encrypted biometric recognition result to the first party who decrypts it to obtain a decrypted biometric recognition result.
2. The system of claim 1 , wherein said homomorphic encryption system is configured to operate both with binary and non-binary plaintexts.
3. The system of claim 2 , wherein said biometric recognition includes a verification process.
4. The system of claim 3 , wherein said input quantization system is configured for nonlinear quantization.
5. The system of claim 4 , wherein said homomorphic encryption system is configured to operate on biometric signals representing faces implementing a secure face recognition system.
6. The system of claim 5 , wherein said biometric signals comprise vectors of Gabor coefficients.
7. The system of claim 6 , wherein said verification process includes a polynomial function.