基于邻域相关性MC模型的空域隐写检测算法

A Detection Algorithm for Spatial-Domain Stego Based on MC Model of Neighborhood Correlation

空域隐写作为最常用的隐写方式之一,对其进行隐写分析具有较高的实用价值。空域隐写破坏了图像的邻域相关性,通过分析其在隐写前后发生的改变,能有效检测图像是否进行了隐写。如何对邻域相关性进行量化分析成为问题解决的关键。通过对图像邻域相关性进行MC建模,提出了一种有效的空域隐写检测算法。该算法在相关性MC模型的基础上提取出相关性量化矩阵,分析了其在隐写前后发生的变化,并提取出相应的检测特征。最后,以Fisher分类器作为检测分类器进行了检测实验。实验结果表明该算法能有效检测多种空域隐写算法,其平均检测率达到85%以上。与其它空域隐写检测算法相比该算法具有更好的适用性。

Spatial - domain stego is a favorite steganography, which of steganalysis has very high applied value. Spatial - domain stego destroys the neighborhood correlation of images, it can detect whether the secret message is hidden in the images to analyze its variance before and after stego. The key problem is how to quantize the neighborhood correlation. A new detection algorithm for spatial - domain stego is proposed by making the MC model of neighborhood correlation. On the basis of MC model of neighborhood correlation, it extracts the corresponding detection features on the basis of analyzing the variance of quantized matrix before and after stego. Finally, make the attack experiments by using Fisher classifier. Experimental results prove that this algorithm can detect various spatial - domain steganographies, its average detection rate is above 85%, and takes on good applicability.