基于同态加密的人脸识别隐私保护方法

Face Recognition Privacy Protection Method Based on Homomorphic Encryption

随着大数据的发展与应用,生物特征识别技术得到了快速发展,并在新型认证技术中开始得到广泛应用.由于传统的基于生物特征的身份认证多是在明文状态下进行,对用户的隐私无法给与充分保障,因此基于以上缺陷提出并设计了一种基于同态加密技术的人脸识别隐私保护方法.该方法首先利用当前热门身份认证模型FaceNet对用户生物特征信息进行提取,然后借助基于RLWE的同态加密技术对提取的特征信息进行加密,保证生物特征信息在外包给服务器进行距离计算时不会泄露用户的隐私数据,防止服务器窥探用户的行为.同时,在身份认证过程中引入随机数概念,防止非法用户对服务器的重放攻击.实验证明,该方法在密文状态下仍能保证较高的准确率与可行性.

With the development and application of big data,biometric recognition technology has developed rapidly and has been widely used in new authentication technology.Because the traditional biometric-based identity authentication is mostly carried out in plaintext,and the user’s privacy cannot be adequately guaranteed,this paper proposes and designs a face recognition privacy protection method based on homomorphic encryption technology based on the above defects.This method firstly uses the current popular authentication model FaceNet to extract the user’s biometric information,and then encrypts the extracted feature information with the help of RLWE based homomorphic encryption technology to ensure that when the biometric information is outsourced to the server for distance calculation,the user’s private data will not be disclosed and the server will not snoop on the user’s behavior.At the same time,in the process of identity authentication,the concept of random number is introduced to prevent illegal users from replaying attacks on the server.Experiments show that the method can still ensure high accuracy and feasibility in the state of ciphertext.