基于SEAL库的同态密文指纹识别系统设计与实现  

Design and Implementation on Homomorphic Ciphertext Fingerprint Recognition System Based on SEAL Library

作　　者：杨亚涛 张奇林[1,2] 张艳硕 左珮良 YANG Ya-Tao;ZHANG Qi-Lin;ZHANG Yan-Shuo;ZUO Pei-Liang(Department of Electronic and Communication Engineering,Beijing Electronic Science and Technology Institute,Beijing 100070,China;School of Telecommunication Engineering,Xidian University,Xi’an 710071,China)

机构地区：[1]北京电子科技学院电子与通信工程系,北京100070 [2]西安电子科技大学通信工程学院,西安710071

出　　处：《密码学报》2021年第4期616-629,共14页Journal of Cryptologic Research

基　　金：“十三五”国家密码发展基金(MMJJ20170110);“信息安全”、“电子信息工程”国家级一流本科专业建设点项目。

摘　　要：将指纹信息直接以明文形式存储在云端数据库会对个人隐私信息造成一定的威胁,同态加密技术和指纹识别的结合,可以在对个人信息提供隐私保护的同时享有云服务提供的强大计算能力和存储能力.为了解决指纹识别中的身份隐私保护和识别效率问题,本文首次提出了一种基于SEAL库的指纹特征密文识别方案.在指纹识别系统中引入全同态加密运算,把密文状态的指纹特征信息存储于云端数据库,通过云计算克服了密文识别中运算开销过大的难题,高效完成了密文识别,可有效抵抗来自内外部的恶意攻击,保障系统的保密性和识别的准确性.本方案基于全同态加密的SEAL库进行开发,经过测试,平均识别时间为1246.33 ms,相比于2018年研究人员提出的基于SEAL的虹膜特征密文认证系统耗时减少64.26%.本方案和系统在密文指纹识别、隐私保护、身份认证领域具有较好的参考价值和应用前景.Storing fingerprint information directly in a cloud database in the form of plaintext would cause a security threat to personal private information.The combination of homomorphic encryption technology and fingerprint recognition can provide privacy preservation for personal information,while it can enjoy the powerful computing and storage capacity provided by cloud services.In order to solve the problems of fingerprint's identity privacy and recognition efficiency,this paper proposes a fingerprint recognition scheme in ciphertext based on SEAL library by using the homomorphic encryption algorithm.By introducing homomorphic encryption into the fingerprint recognition system and storing the fingerprint characteristic information in ciphertext state in the cloud database,the problem of excessive computing overhead can be overcome by cloud computing in ciphertext recognition system.Moreover,a fingerprint recognition system in ciphertext is designed,which can resist malicious attacks from inside and outside and ensure the confidentiality and accuracy of the system.This scheme is developed based on the homomorphic encryption SEAL library.Experimental tests show that,compared with iris feature's recognition system in ciphertext based on SEAL proposed in 2018,the average recognition time is only 1246.33 ms,which reduces by 64.26%.The proposed scheme can have good applications in ciphertext fingerprint recognition,privacy preservation and identity authentication.

关 键 词：指纹识别 同态加密 SEAL 识别率 认假率 拒真率 

分 类 号：TP309.7[自动化与计算机技术—计算机系统结构]