深度学习在图像隐写术和隐写分析领域中研究进展.PDFVIP

第3 卷 第6 期 信 息 安 全 学 报 Vol. 3 No. 6 2018 年11 月 Journal of Cyber Security November, 2018 深度学习在图像隐写术与隐写分析领域 中的研究进展 翟黎明, 嘉 炬, 任魏翔, 徐一波, 王丽娜 武汉大学 国家网络安全学院 空天信息安全与可信计算教育部重点实验室 武汉 中国 430072 摘要 隐写术与隐写分析是信息安全领域的热门研究方向, 近年来得到了广泛的研究与快速的发展。随着深度学习新技术的兴 起, 深度学习也被引入到隐写术与隐写分析领域, 并在方法和性能上取得了一系列突破性的研究成果。为推进基于深度学习的 隐写术与隐写分析的研究, 本文对目前的主要方法和代表性工作进行了归纳与探讨。对于图像隐写术与隐写分析这两个领域, 本文分别各自比较了传统方法和与相关深度学习方法的异同, 详细介绍了目前主要的基于深度学习的图像隐写术与隐写分析的 基本原理和方法, 最后讨论了基于深度学习的图像隐写术与隐写分析仍需要解决的问题及未来的研究趋势。 关键词 隐写术; 隐写分析; 深度学习; 生成对抗网络; 卷积神经网络 中图法分类号 TP391 DOI 号 10.19363/J.10-1380/tn.2018.11.01 Recent advances in deep learning for image steganography and steganalysis ZHAI Liming, JIA Ju, REN Weixiang, XU Yibo, and WANG Lina Key Laboratory of Aerospace Information Security and Trusted Computing, Ministry of Education, School of Cyber Science and Engineering, Wuhan University, Wuhan 430072, China Abstract steganography and steganalysis are hot research directions in the area of information security, and they have been widely researched and rapidly developed in recent decades. With the rise of new technologies for deep learning, steganography and steganalysis based on deep learning have achieved a series of breakthrough in methods and perform- ance. In order to promote the research of steganography and steganalysis based on deep learning, typical methods and rep- resentative work are summarized and discussed in this paper. For image steganography and steganalysis, the similarities and differences between conventional methods and deep learning based methods are compared respectively, and the basic principles an