计算机工程与应用 ›› 2022, Vol. 58 ›› Issue (11): 93-99.DOI: 10.3778/j.issn.1002-8331.2112-0517

• 网络、通信与安全 • 上一篇    下一篇

FaceEncAuth:基于FaceNet和国密算法的人脸识别隐私安全方案

吴俊青,彭长根,谭伟杰,吴振强   

  1. 1.贵州大学 计算机科学与技术学院,贵阳 550025
    2.贵州大学 公共大数据国家重点实验室,贵阳 550025
    3.贵州大学 贵州省大数据产业发展应用研究院,贵阳 550025
    4.陕西师范大学 计算机科学学院,西安 710062
  • 出版日期:2022-06-01 发布日期:2022-06-01

FaceEncAuth:Face Recognition Privacy Security Scheme Based on FaceNet and SM Algorithms

WU Junqing, PENG Changgen, TAN Weijie, WU Zhenqiang   

  1. 1.College of Computer Science and Technology, Guizhou University, Guiyang 550025, China
    2.State Key Laboratory of Public Big Data, Guizhou University, Guiyang 550025, China
    3.Guizhou Big Data Academy, Guizhou University, Guiyang 550025, China
    4.College of Computer Science, Shaanxi Normal University, Xi’an 710062, China
  • Online:2022-06-01 Published:2022-06-01

摘要: 人脸识别中,人脸特征作为生物特征的一种,具有唯一性、不可撤销性,一旦遭到攻击、篡改或泄露,用户隐私安全将面临巨大威胁。针对这一问题,提出一种基于深度学习和加密算法的人脸识别隐私安全方案。该方案中,利用FaceNet深度学习算法来高效提取人脸特征,协调生物特征模糊性与密码系统的精确性,采用CKKS全同态加密算法进行人脸识别密文域的运算,通过国密SM4算法增强人脸特征密文抵抗恶意攻击的能力,利用其对称密码的性质兼顾了安全性和运算效率,而SM9非对称密码算法则用于SM4算法对称密钥的管理。实验结果及分析表明,该方案在不影响人脸识别准确率、效率的前提下提高了数据传输、存储和比对的安全性。

关键词: 人脸识别, 生物特征, 深度学习, 同态加密, SM4, SM9

Abstract: In face recognition, as one of the basic biometrics, facial features are unique and irrevocable. The attack, leakage or the tampering with facial features will pose a great threat to user privacy and security. To solve this problem, a face recognition privacy security scheme based on deep learning and encryption algorithms is proposed. In this scheme, the FaceNet deep learning algorithm is used to efficiently extract facial features, coordinate the ambiguity of biometrics and the accuracy of the cryptographic system, and use the CKKS fully homomorphic encryption algorithm for face recognition ciphertext field operations. Besides, the national secret SM4 algorithm is adopted to enhance facial feature ciphertext’s ability to resist malicious attacks, in which symmetric cipher property is utilized to take into account both security and operation efficiency, while the SM9 asymmetric cipher algorithm is used for the management of the symmetric key of the SM4 algorithm. Experimental results and analysis show that the scheme improves the security of data transmission, storage, and comparison without affecting the accuracy and efficiency of face recognition.

Key words: face recognition, biometrics, deep learning, homomorphic encryption, SM4, SM9