计算机应用 ›› 2009, Vol. 29 ›› Issue (09): 2344-2347.

• 信息安全 • 上一篇    下一篇

基于DCT系数多方向相关性的信息隐藏盲检测方法

王勇1,刘九芬2,张卫明2   

  1. 1. 河南省郑州市金水区俭学街信息工程学院四系研究生队
    2.
  • 收稿日期:2009-03-12 修回日期:2009-04-20 发布日期:2009-11-10 出版日期:2009-09-01
  • 通讯作者: 王勇
  • 基金资助:
    国家自然科学基金(NO.60803155);国家级基金

Blind image steganalysis based on multi-directional correlation of DCT coefficient

  • Received:2009-03-12 Revised:2009-04-20 Online:2009-11-10 Published:2009-09-01

摘要: 给出了一种JPEG域基于离散余弦变换(DCT)系数多方向相关性的图像信息隐藏盲检测方法。首先利用DCT系数多方向相关性构造差值彼邻相关矩阵,然后利用此矩阵提取48维特征向量,最后使用支持向量机(SVM)区分载体、载密图像。该方法应用于6种典型的JPEG图像隐写算法,实验结果表明,该方法对这些隐写算法都能够进行比较可靠的检测。

关键词: 信息隐藏, 盲检测, 差值彼邻相关矩阵, 支持向量机

Abstract: A blind steganalysis method based on multi-orientation correlation of Discrete Cosine Transform (DCT) coefficient was presented. This method first applied multi-directional correlation of DCT coefficient to construct difference neighboring correlation matrix, then extracted 48 dimensional features for each image, finally Support Vector Machine (SVM) was used to classify the cover and stego images. A series of experiments were performed on 6 kinds of typical steganography of different embedding ratio. The results show that this method can make a reliable blind detection for these typical steganographic schemes.

Key words: information hiding, blind detection, difference neighboring correlation matrix, Support Vector Machine (SVM)

中图分类号: