Research on Privacy Protection of Face Recognition Based on Function Encryption
As a kind of biometric technology,face recognition technology has been widely used in daily life,and each person's unique face information will become a unique identity proof.However,there're also a series of prob-lems with facial recognition technology.At present,in many authentication occasions,face recognition usually does not have a complete encryption of the stored face information,and personal biological information is likely to be leaked,which will cause irreversible serious consequences.In this paper,we propose a function-encrypted face recog-nition system scheme,combined with the FHIPE encryption scheme in the face recognition system,based on the ex-isting OpenFace and other feature extraction libraries,to ensure efficient recognition and protect the user's face priva-cy.When we do the user's face authentication,it only pays attention to the inner product of the encrypted vectors,and does not expose the original face feature vector.Even if the data is leaked,the face features cannot be extracted from encrypted vectors which ensures the security of the system.In this paper,9 volunteers were invited to participate in the experiment to verify the effectiveness and robustness of the user's face recognition under different encryption vector digits.The results indicate that the larger the number of digits of the encryption vectors,the longer the authenti-cation time,but the better the effect of face authentication.When the number of digits of the encryption vector is 32,the authentication time is 3.19 seconds,and the false acceptance rate(FAR)and false rejection rate(FRR)are 0.099%and 0.22%,respectively.
Face recognitionUser privacy protectionFace information encryption
NETL
NSTL
万方数据
维普
