基于GAN的生成式图像隐写方案  

Generative Image Steganography Scheme Based on GANs

作　　者：王耀杰 杨晓元[1,2] 刘佳[1,2] WANG Yao-jie;YANG Xiao-yuan;LIU Jia(Key Laboratory for Network and Information Security of Chinese Armed Police Force,Engineering University of PAP,Xi'an 710086,China;College of Cryptographic Engineering, Engineering College of PAP, Xi'an 710086,China)

机构地区：[1]武警工程大学网络与信息安全武警部队重点实验室,西安710086 [2]武警工程大学密码工程学院,西安710086

出　　处：《小型微型计算机系统》2019年第8期1740-1744,共5页Journal of Chinese Computer Systems

基　　金：国家重点研发计划项目(2017YFB0802000)资助;国家自然科学基金项目(61379152,61403417)资助

摘　　要：针对信息隐藏中载体修改会留有修改痕迹、载体选择困难且负载量太低的问题,本文在载体合成的基础上提出了一种基于GAN的生成式图像隐写方案(GS-GAN). GS-GAN在生成对抗网络的基础上,采用隐藏的信息作为驱动直接生成含密图像进行传输,符合"无载体"信息隐藏思想,可有效抵抗基于统计特性的隐写检测.同时,基于加密算法的密钥,即使截获隐藏的内容,在无密钥的前提下,只会获得无意义的结果.我们在CelebA数据集进行了实验,验证了该方案的安全优势和可行性.According to the embedding method of secret information,steganography can be divided into: cover modification,selection and synthesis. Aiming at the problem that the cover modification in the information hiding will leave the modification trace,the cover selection is difficult and the load is too low,combined with the idea of cover synthesis,a generative steganography scheme based on GANs is proposed.( GS-GAN). On the one hand,the GAN will use the information to be hidden as a driver to directly generate the secret. in line with coverless information hiding,can effectively resist steganalysis based on statistical analysis algorithm of detection,On the other hand,the scheme is based on the private key of symmetric cryptography. Even if the hidden contents are intercepted,only meaningless results can be obtained without the secret key. By GAN will stay hidden information. We conducted experiments on the CelebA dataset,which verified the feasibility and safety of the method.

关 键 词：信息隐藏 对称加密 生成式 生成对抗网络 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]