通信学报 ›› 2020, Vol. 41 ›› Issue (1): 125-133.doi: 10.11959/j.issn.1000-436x.2020019

• 学术论文 • 上一篇    下一篇

DGANS:基于双重生成式对抗网络的稳健图像隐写模型

竺乐庆1,郭钰1,莫凌强1,张大兴2   

  1. 1 浙江工商大学计算机与信息工程学院,浙江 杭州 310018
    2 杭州电子科技大学计算机学院,浙江 杭州 310018
  • 修回日期:2019-12-05 出版日期:2020-01-25 发布日期:2020-02-11
  • 作者简介:竺乐庆(1972- ),女,浙江嵊州人,博士,浙江工商大学副教授、硕士生导师,主要研究方向为图像处理、模式识别、视频处理、信息隐藏等|郭钰(1995- ),男,安徽宿州人,浙江工商大学硕士生,主要研究方向为图像处理、模式识别、图像信息隐藏等|莫凌强(1994- ),男,浙江嘉兴人,浙江工商大学硕士生,主要研究方向为模式识别、图像处理、图像信息隐藏等|张大兴(1971- ),男,浙江嵊州人,博士,杭州电子科技大学副教授、硕士生导师,主要研究方向为信息安全、多媒体技术、软件工程等
  • 基金资助:
    国家自然科学基金资助项目(61572160);浙江省自然科学基金资助项目(LY20F020002)

DGANS:robustness image steganography model based on double GAN

Leqing ZHU1,Yu GUO1,Lingqiang MO1,Daxing ZHANG2   

  1. 1 School of Computer and Information Engineering,Zhejiang Gongshang University,Hangzhou 310018,China
    2 School of Computer Science and Technology,Hangzhou Dianzi University,Hangzhou 310018,China
  • Revised:2019-12-05 Online:2020-01-25 Published:2020-02-11
  • Supported by:
    The National Natural Science Foundation of China(61572160);The Natural Science Foundation of Zhejiang Province(LY20F020002)

摘要:

深度卷积神经网络可有效地应用于大容量图像信息隐写,然而其稳健性研究却鲜有报道。双重生成式对抗网络(DGANS)模型对深度学习框架应用于图像隐写时,针对小幅度几何变换攻击进行了优化设计,从而提高模型的稳健性。DGANS由2个串联的生成式对抗网络构成,可将灰度图像隐藏到相同大小的彩色或灰度图像中并还原。通过对生成的含密图像进行数据增强并进一步强化训练提取网络,使提取网络对输入图像的几何变换具有适应性。实验结果表明,DGAN不仅可以实现高容量的图像信息隐写,而且可以对抗一定范围内的几何攻击,比同类模型有更好的稳健性。

关键词: 图像隐写, 稳健性, 双重生成式对抗网络, 深度学习

Abstract:

Deep convolutional neural networks can be effectively applied to large-capacity image steganography,but the research on their robustness is rarely reported.The DGANS (double-GAN-based steganography) applies the deep learning framework in image steganography,which is optimized to resist small geometric distortions so as to improve the model’s robustness.DGANS is made up of two consecutive generative adversarial networks that can hide a grayscale image into another color or grayscale image of the same size and can restore it later.The generated stego-images are augmented and used to further train and strengthen the reveal network so as to make it adaptive to small geometric distortion of input images.Experimental results suggest that DGANS can not only realize high-capacity image steganography,but also can resist geometric attacks within certain range,which demonstrates better robustness than similar models.

Key words: image steganography, robustness, double GAN, deep learning

中图分类号: 

No Suggested Reading articles found!