• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>IFIP TC-6 TC-11 International Conference on Communications and Multimedia Security(CMS 2005); 20050919-21; Salzburg(AT) >Blind Statistical Steganalysis of Additive Steganography Using Wavelet Higher Order Statistics
【24h】

Blind Statistical Steganalysis of Additive Steganography Using Wavelet Higher Order Statistics

机译：利用小波高阶统计进行附加隐写术的盲统计隐写分析

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Development of digital communications systems significantly extended possibility to perform covert communications (steganography). This recalls an emerging demand in highly efficient counter-measures, i.e. Steganalysis methods. Modern steganography is presented by a broad spectrum of various data-hiding techniques. Therefore development of corresponding Steganalysis methods is rather a complex problem and challenging task. Moreover, in many practical Steganalysis tasks second Kerckhoff's principle is not applicable because of absence of information about the used steganography method. This motivates to use blind Steganalysis, which can be applied to the certain techniques where one can specify at least statistics of the hidden data. This paper focuses on the class of supervised Steganalysis techniques developed for the additive steganography, which can be described as y = f(x, s, K) = x + g(s, K), where stego image y is obtained from the cover image x by adding a low-amplitude cover image independent (±1 embedding also known as LSB matching) or cover image dependent (LSB embedding) stego signals that may be also depended on secret stego key K and the secret data s. The function g(.) represents the embedding rule. The proposed method provides the stochastic interpretation of the blind Steganalysis and consists of two main stages, i.e., data preprocessing and feature extraction. The data preprocessing targets at stego signal estimation that is performed in the wavelet domain from the mixture of cover image (presented by non-stationary Gaussian model) and stego signal (presented by stationary Gaussian model). Feature extraction is realized using model-based (polynomial) approximation of stego image pdf. In this case polynomial coefficients, which simultaneously are high order statistics, have created the feature set. Because the features are calculated from the estimated stego signal, they are more sensitive to steganographic modifications while suppressing the influence of the cover image.
机译：数字通信系统的发展极大地扩展了进行秘密通信（隐写术）的可能性。这回想起了对高效应对措施（即隐写分析方法）的新兴需求。现代隐写术是由各种各样的各种数据隐藏技术提供的。因此，开发相应的隐写分析方法是一个复杂的问题，也是一项艰巨的任务。此外，在许多实际的隐写分析任务中，由于缺少有关所用隐写术方法的信息，因此第二Kerckhoff原理不适用。这激发了使用盲隐写分析的能力，该隐写隐写分析可以应用于至少可以指定隐藏数据统计信息的某些技术。本文着重介绍了针对加法隐写技术开发的有监督隐写分析技术，可以将其描述为y = f（x，s，K）= x + g（s，K），其中隐秘图像y是从封面获得的通过添加独立于低幅覆盖图像（±1嵌入，也称为LSB匹配）或依赖覆盖图像（LSB嵌入）的隐身信号（也可能取决于秘密隐身密钥K和秘密数据s）添加图像x。函数g（。）表示嵌入规则。所提出的方法提供了对盲隐写分析的随机解释，并包括两个主要阶段，即数据预处理和特征提取。数据预处理的目标是根据覆盖图像（由非平稳高斯模型表示）和隐秘信号（由固定高斯模型表示）的混合信号在小波域中进行隐秘信号估计。使用基于模型的（隐式）隐身图像pdf近似实现特征提取。在这种情况下，同时是高阶统计量的多项式系数已经创建了特征集。因为这些特征是根据估计的隐身信号计算得出的，所以它们在对隐匿性修改更敏感的同时抑制了封面图像的影响。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号