对抗性增强的图像块位平面拆分缩略图保留加密  

Resistance-enhanced image block bit-plane split thumbnail preserving encryption

作　　者：李名[1,2] 崔清晨 王曦 张静 李文泽 Li Ming;Cui Qingchen;Wang Xi;Zhang Jing;Li Wenze(College of Computer and Information Engineering,Henan Normal University,Xinxiang 453007,China;Key Laboratory of Artificial Intelligence and Personalized Learning in Education of Henan Province,Henan Normal University,Xinxiang 453007,China)

机构地区：[1]河南师范大学计算机与信息工程学院,河南新乡453007 [2]河南师范大学河南省教育人工智能与个性化学习重点实验室,河南新乡453007

出　　处：《河南师范大学学报（自然科学版）》2025年第1期92-99,共8页Journal of Henan Normal University(Natural Science Edition)

基　　金：国家自然科学基金(61602158);河南省高等学校重点科研项目(23A520009).

摘　　要：随着个人图像数量日益增加,云服务开始在图像存储方面发挥重要作用.然而,将图像上传到云端将会面临隐私威胁.传统的加密方案对图像进行简单加密就可以保护图像隐私,但它牺牲了图像内容的可用性.近年来,缩略图保留加密(thumbnail preserving encryption,TPE)被提出,通过加密后保持缩略图不变使云中图像在不被非法第三方肉眼识别的同时能对用户具有可用性.但是现有的TPE方案没有考虑到机器学习对图像的隐私威胁.基于此,提出了一种新的TPE方案,该方案在加密过程中从对抗深度学习模型识别图像这一全新的角度出发,提高了保护图像隐私信息的能力.实验表明,所提出的方案能够在图像对用户具有可用性的同时抵抗人类肉眼和深度神经网络识别图像.With the increasing number of personal images,cloud services are beginning to play an important role in image storage.However,uploading images to the cloud leads to privacy threats.Traditional encryption schemes that simply encrypt images can protect image privacy,but they sacrifice the usability of the image content.In recent years,thumbnail preserving encryption(TPE)has been proposed to make images available to users in the cloud without being identified by illegal third parties with the unchanged thumbnails after encryption.However,existing TPE schemes do not take account of the privacy threat to images from machine learning.We propose a new TPE scheme that improves the ability of the algorithm to protect image privacy information from a new perspective by combating deep learning models in the encryption process.Experiments show that the proposed scheme is able to resist image recognition by human eyes and deep neural networks while the image is usable.

关 键 词：图像加密 对抗性 可用性 隐私性 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]