摘要
传统的隐写方式面临的威胁越来越大,隐写分析技术也逐渐成熟,针对这一问题,将生成式对抗网络引入隐写术中,可以减少载体修改痕迹,提高隐写的隐蔽性.介绍了生成式对抗网络的基本结构,总结了基于GAN图像生成的隐写技术的研究成果,并进行比较和分类.根据已有的技术手段提出了当前生成式对抗网络在隐写技术发展中的不足,对未来的研究方向进行了展望.
The traditional steganography is facing more and more threats, and the steganographic analysis technology is gradually mature. To solve this problem, the Generative Adversarial Networks is introduced into the steganography, which can reduce the traces of carrier modification and improve the concealment of steganography. This paper introduces the basic structure of the Generative Adversarial Networks, summarizes, compares and classifies the research results of the steganography based on GAN image generation. According to the existing technical means, the shortcomings of the current Generative Adversarial Networks in the development of steganography are proposed, and the future research directions are prospected.
作者
周琳娜
吕欣一
Zhou Linna;Lü Xinyi(School of Information Science and Technology,University of International Relations,Beijing 100091)
出处
《信息安全研究》
2019年第9期771-777,共7页
Journal of Information Security Research
基金
国家自然科学基金重点项目(U1536207)
国家重点研发计划项目(2016QY08D1600)
国家重点研发计划课题(2016YFB0801405)
2018年度中央高校基本科研业务费项目(3262018T02)
关键词
生成式对抗网络
信息隐藏
隐写术
深度学习
图像生成
generative adversarial networks
information hiding
steganography
deep learning
image generation
