东北大学学报(自然科学版) ›› 2021, Vol. 42 ›› Issue (8): 1079-1085.DOI: 10.12068/j.issn.1005-3026.2021.08.003

• 信息与控制 • 上一篇    下一篇

具有前向安全特性的密态图像搜索方案

葛悦1, 李宇溪1, 周福才1, 张鑫月2   

  1. (1. 东北大学 软件学院, 辽宁 沈阳110169; 2. 中国移动通信有限公司研究院, 北京100032)
  • 修回日期:2020-11-30 接受日期:2020-11-30 发布日期:2021-09-02
  • 通讯作者: 葛悦
  • 作者简介:葛悦(1995-),女,河南焦作人,东北大学硕士研究生; 周福才(1964-),男,辽宁沈阳人,东北大学教授,博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(62072090); 中央高校基本科研业务费专项资金资助项目(N171704005).

Encrypted Image Search Scheme with Forward Security

GE Yue1, LI Yu-xi1, ZHOU Fu-cai1, ZHANG Xin-yue2   

  1. 1. School of Software, Northeastern University, Shenyang 110169, China; 2. China Mobile Research Institute, Beijing 100032, China.
  • Revised:2020-11-30 Accepted:2020-11-30 Published:2021-09-02
  • Contact: ZHOU Fu-cai
  • About author:-
  • Supported by:
    -

摘要: 围绕当前多数密态图像搜索方案在更新时产生的信息泄露问题,提出一个云环境下具有前向安全特性的密态图像搜索方案.方案首先采用Tensorflow机器学习模型进行图像识别,设计高效图像关键字信息提取方法;然后基于动态可搜索加密思想,利用GGM范围受限伪随机函数生成搜索令牌,构造出支持动态更新的安全密态图像索引结构.搜索时利用搜索令牌和哈希函数计算匹配结果,不会泄露查询者隐私.安全分析和实验结果表明,本方案在满足高效搜索的同时实现前向安全性,并具有良好的更新性能.

关键词: 加密图像;动态可搜索加密;前向安全;受限伪随机函数;降采样预处理

Abstract: In most encrypted image search schemes, the information leaked during the update is easily used by attackers. In order to solve this problem, an encrypted image search scheme with forward security is proposed. A Tensorflow machine learning model implements image recognition and an efficient image keyword extraction method is designed, and based on dynamic searchable encryption, a GGM range-constrained pseudorandom function constructs search tokens and a secure encrypted image index that supports dynamic update. In the search phase, the search token and Hash function are used to calculate the matching results, and the privacy of the query is not disclosed. Security analysis and experimental results show that this scheme achieves forward security and guarantees good update performance.

Key words: encrypted image; dynamic searchable encryption; forward security; constrained pseudorandom function; down sampling preprocessing

中图分类号: