面向秘密共享的逐层残差预测加密域大容量数据隐藏

温文媖; 杨育衡; 张玉书; 方玉明; 邱宝林

引用本文：

温文媖,杨育衡,张玉书,方玉明,邱宝林.面向秘密共享的逐层残差预测加密域大容量数据隐藏[J].信息安全学报,2025,10(1):61-74 [点击复制]
WEN Wenying,YANG Yuheng,ZHANG Yushu,FANG Yuming,QIU Baolin.High-Capacity Data Hiding in Encryption Domain Based on Layer-by-Layer Residual Prediction for Secret Sharing[J].Journal of Cyber Security,2025,10(1):61-74 [点击复制]

本文已被：浏览 99次下载 49次	码上扫一扫！
面向秘密共享的逐层残差预测加密域大容量数据隐藏
温文媖¹, 杨育衡¹, 张玉书^2,3, 方玉明¹, 邱宝林¹
0 字体:加大+\|默认\|缩小-
(1.江西财经大学信息管理学院南昌中国 330032;2.南京航天航空大学计算机科学与技术学院南京中国 210016;3.中国科学院信息工程研究所信息安全国家重点实验室北京中国 100093)

摘要:

加密图像中的数据隐藏(Data Hiding in Encrypted Images,DHEI)是一种可行的云端存储方案,但其载体唯一,一旦被破坏就可能导致载体图像无法恢复。DHEI与秘密共享的结合能够在多载体图像中嵌入数据的同时保护原始图像的隐私性和安全性。但现有基于数据隐藏的秘密共享方案主要是利用自然图像像素的相关性为数据隐藏预留空间,嵌入容量受自然图像内容制约。在进行数据嵌入时,若数据量大于载体图像可嵌入容量,则存在数据丢失的可能性。针对该问题,本文基于压缩感知技术(Compressed Sensing,CS),提出一种面向秘密共享的逐层残差预测加密域大容量数据隐藏方案。首先,该方案通过压缩感知逐层预测技术(Layer-by-Layer Prediction Technology base on Compressed Sensing,LLPT-CS)减小测量值之间的冗余性,实现对原始图像进行加密的同时腾出嵌入空间(~4.0bpp);其次,加密图像以秘密图像共享(Secret Image Sharing,SIS)的形式生成n个秘密份额,分别发送至n个数据隐藏器;接着,数据隐藏器在无图像内容访问权限的情况下向秘密份额嵌入秘密数据;最后,接收端获取n个数据隐藏器中的任意k个秘密份额后即可依次通过拉格朗日插值法和CS重建算法恢复原始图像。实验结果表明,本文提出方案能实现嵌入率预设,保证数据嵌入的稳定性,并且能较好地保护云端图像存储的隐私性和安全性;与现有的秘密共享数据隐藏方案相比,该方案不仅能很好地为云端图像储存提供稳定的大容量秘密数据嵌入空间,而且还能恢复出在视觉上愉悦的图像,拥有现有方案不具备的逐步恢复功能。

关键词: 逐层残差预测压缩感知数据隐藏秘密共享

DOI：10.19363/J.cnki.cn10-1380/tn.2025.01.05

投稿时间：2023-02-13修订日期：2023-06-20

基金项目:本课题得到国家自然科学基金(No.62201233,No.61961022)资助、国家信息安全国家重点实验室基金(No.2022-MS-02)、江西省双千计划(No.jxsq2023201118)、江西省杰出青年基金项目(No.20232ACB212004)、江西省教育厅基金(No.GJJ210502)资助。

High-Capacity Data Hiding in Encryption Domain Based on Layer-by-Layer Residual Prediction for Secret Sharing

WEN Wenying¹, YANG Yuheng¹, ZHANG Yushu^2,3, FANG Yuming¹, QIU Baolin¹

(1.School of Information Technology, Jiangxi University of Finance and Economics, Nanchang 330032, China;2.College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China;3.State Key Laboratory of Information Security, Institute of Information Engineering, Chinese Academy of Sciences, Beijing 100093, China)

Abstract:

Data Hiding in Encrypted Images (DHEI) is a feasible cloud storage scheme, but its carrier image is unique, and once it is destroyed, the carrier image may be unrecoverable. The combination of data hiding and secret sharing in the encrypted domain can embed data in multiple carrier images while preserving the privacy and security of the original image. However, current secret-sharing schemes based on data hiding primarily employ the correlation of natural image pixels to reserve space for data hiding, and the embedding rate is limited by the natural image content. When embedding data, if the amount of data is greater than the embedding capacity of the carrier image, there is a possibility of data loss. To address this problem, based on compressed sensing(CS), this paper proposes a high-capacity data hiding scheme in encryption domain by employing layer-by-layer residual prediction for secret sharing scheme for secret sharing. First of all, the scheme uses layer-by-layer prediction technology base on compressed sensing(LLPT-CS) to reduce the redundancy between measured values, which can encrypt the original image while freeing up the embedding space(~4.0bpp). Second, the encrypted image creates n secret shares in the form of secret image sharing(SIS) and sends them to each of the n data hiders. Next, the data hiders embed secret data into the secret shares without having access to the image content. In the end, the receiver receives any of the k secret shares from n data hiders and can then recover the original image using Lagrangian interpolation and CS reconstruction algorithms. The experimental findings demonstrate that the proposed scheme in this paper provides more stable and large-capacity secret data embedding space for cloud image storage schemes than existing secret sharing data hiding schemes. Additionally, the proposed scheme recovers aesthetically pleasing images with a progressive recovery function that existing schemes lack.

Key words: layer-by-layer residual prediction compressed sensing data hiding secret sharing