华南理工大学学报(自然科学版) ›› 2008, Vol. 36 ›› Issue (4): 86-92.

• 计算机科学与技术 • 上一篇    下一篇

隐写术安全性度量模型及其应用

何军辉 唐韶华   

  1. 华南理工大学 计算机科学与工程学院, 广东 广州 510640
  • 收稿日期:2007-05-30 修回日期:2007-09-04 出版日期:2008-04-25 发布日期:2008-04-25
  • 通信作者: 何军辉(1976-),男,讲师,博士,主要从事信息安全研究. E-mail:hejh@scut.edu.cn
  • 作者简介:何军辉(1976-),男,讲师,博士,主要从事信息安全研究.
  • 基金资助:

    广东省信息安全技术重点实验室2006年度开放基金资助项目

Evaluation Model of Steganography Security and Its Application

He Jun-hui  Tang Shao-hua   

  1. School of Computer Science and Engineering, South China University of Technology, Guangzhou 510640, Guangdong, China
  • Received:2007-05-30 Revised:2007-09-04 Online:2008-04-25 Published:2008-04-25
  • Contact: 何军辉(1976-),男,讲师,博士,主要从事信息安全研究. E-mail:hejh@scut.edu.cn
  • About author:何军辉(1976-),男,讲师,博士,主要从事信息安全研究.
  • Supported by:

    广东省信息安全技术重点实验室2006年度开放基金资助项目

摘要: 关于隐写术安全性的度量,目前尚缺乏有效的模型.文中在建立图像离散余弦变换(DCT)系数统计分布模型的基础上,根据隐写系统安全性的信息论定义,推导和建立了一种DCT域上的隐写术安全性度量模型.利用此模型对常见空域和DCT域隐写术的安全性进行了对比分析,同时进行了模型求解方法、隐藏容量、分块大小、嵌入策略和两次压缩等因素对安全性度量影响的实验.结果表明,文中提出的模型是有效的,能为隐写术和隐写分析算法的设计提供有价值的参考.

关键词: 隐写术, 度量模型, 离散余弦变换, 高斯分布, 最大似然估计, 鉴别信息

Abstract:

At present, there are few effective models to evaluate the security of steganography. In this paper, first, a model to describe the statistical distribution of the discrete cosine transform (DCT) coefficients of an image is es- tablished, and a model based on the information-theory definition of steganography security is presented to measure the security of steganography in the DCT domain. The securities of several steganographic techniques in both spatial and DCT domains are then compared. Moreover, some experiments are performed to measure the effects of the methods of model solving, the size of hiding message, the block size in DCT, the embedding position and the dou- ble compression on the security evaluation. The results show that the proposed model is an effective model, which provides valuable references for the design of steganography and steganalytic algorithms.

Key words: steganography, evaluation model, discrete cosine transform, Gaussian distribution, maximum likelihood estimation, discrimination information