基于局部线性变换的HUGO图像隐写检测  

Based on Local Linear Transformation HUGO Image Steganography Detection

作　　者：邓裕[1] 袁晓红[1] 秦姣华[1] 

机构地区：[1]中南林业科技大学计算机与信息工程学院

出　　处：《电子世界》2014年第1期100-102,共3页Electronics World

基　　金：国家自然科学基金资助项目(项目编号:61202496);湖南省自然科学基金重点资助项目(项目编号:13JJ2031)

摘　　要：Pevny等人2010年提出一种最先进的高度不可检测的自适应空域隐写算法HUGO(Highly Undetectable Steganography),其保护了相邻四个像素的一阶统计特性,具有很强的抗检测性能,目前国内外针对HUGO检测缺乏有效的方法。由于用HUGO隐写算法进行隐写嵌入时改变部分纹理特征,本文提出一种新的检测方法,利用局部线性变换得到纹理残差图像,计算共生矩阵得到22130维特征向量,最后使用集成分类器进行分类。实验结果表明:在嵌入率为0.4bpp时,针对BOSSRank图像集,获得平均82.71%的检测率,优于Hugobreaker的80.3%和Guel的76.8%,并在低于0.4bpp嵌入率时,其检测效果有所提高。Since Pevny put forward a kind of the most advanced Highly Undetectablc Steganography in 2010 , which approximately preserves the joint distribution of the first-order differences between four neighboring pixels and has a strong resistance to detection performance.Now there is a lack of effective methods in HUGO detection at home and abroad. Steganographic embedding with HUGO changed parts of the textural feature, this paper proposes a new detection method, which is first obtained textural residual image by local linear transformation, m get 22130 dimensionality co-occurrence matrix feature vector, finally using the ensemble classifier to classify.The experimental results show that when the embedding rate is 0.4 bit per pixel, we obtain a detection rate of 82.71% on average using BOSSRank Image Sets, which is better than 80.3% of Hugobreakers, 76.8% of Guel, also in less than 0.4 bit per pixel, the detection effect also improved.

关 键 词：隐写分析 HUGO 残差图像 共生矩阵 集成分类器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]