针对LSB匹配隐写的图像复原隐写分析  

Image Restoration-Based Steganalysis Directed to LSB Matching Steganography

作　　者：徐旭[1] 平西建[1] 张涛[1] 王国新[1] 

机构地区：[1]信息工程大学信息工程学院,郑州450002

出　　处：《计算机辅助设计与图形学学报》2009年第2期262-267,274,共7页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(60473022)

摘　　要：最低有效位(LSB)匹配隐写是目前图像隐写分析的难点和热点之一.为了提高针对LSB匹配隐写的隐写分析算法性能,将图像退化/复原理论与图像隐写分析相结合,提出一种新的隐写分析算法.首先将LSB匹配隐写过程建模为加性噪声造成的图像退化过程,提出了一种专用复原滤波算法;然后将载密图像的复原图像作为载体图像的估计图像,提取载密图像与估计图像的质心特征,结合Fisher线性判决器实现隐写分析.实验结果表明,复原滤波算法可有效地复原受LSB匹配隐写噪声污染的退化图像,隐写分析算法的总体性能优于Ker方法,尤其在低嵌入率条件下表现良好,适用于空间域图像LSB匹配隐写.The least significant bit（LSB） matching steganography is one of the most difficult and attractive subjects to steganalysts. To achieve better detection performance, this paper proposes a steganalysis algorithm applied to spatial image, by combing the image degradation/restoration theory with steganalysis. By the algorithm, firstly we model the image degradation process caused by additive noise, on embedding process, and proposes a spatial restoration filter correspondingly. In case the restored image of stego-image is taken as the estimation of the cover-image, we calculate the center-of-mass feature vector between the original image and the restored one. And the steganalysis is accomplished by introducing the mentioned feature vector into Fisher linear discrimination. Experimental results show that the proposed filter manages to recover the stego-degraded image, and the steganalysis method generally performs better than Ker＇s method, especially in the condition of low embedding rate. The proposed steganalysis algorithm is applicable to LSB matching steganography for spatial images.

关 键 词：信息隐藏 隐写分析 图像退化/复原 LSB匹配隐写 Lee滤波 

分 类 号：TN911.73[电子电信—通信与信息系统]