基于像素差分相关性的LSB匹配隐写分析

针对空间域LSB匹配隐写,提出了一种基于像素差分相关性的隐写分析方法.采用图像的直方图、差分直方图的平滑度、梯度能量、图像平滑度,像素差分的一维和高维统计分布等作为分类特征,结合Fisher线性判别进行隐写分析.实验结果表明,新算法对于LSB匹配的检测具有较高的准确性,且计算复杂度低.该算法还可应用于其他空间域隐写方法的检测.

Steganalysis of LSB Matching Exploiting Correlations Between Pixel Differences

In this paper,a new steganalytic method exploiting correlations between pixel differences is proposed,which is designed to detect the presence of spatial LSB matching steganography in grayscale images.In the proposed method,the smoothness of the image histogram and the pixel difference histogram,the gradient energy,the image smoothness,the one-dimensional and high-dimensional statistical distributions of pixel differences are used as distinguishing features,and the Fisher linear discrimination analysis is used to construct a classier to do the steganalytic job.Experimental results show that the proposed method exhibits excellent performances for the detection of LSB matching steganography in grayscale images.Moreover,it has a low computational complexity.The proposed steganalytic method can also be applied to the detection of other spatial steganography methods.