基于多特征的F5隐写分析算法

通过分析F5算法给载体图像带来的统计变化,从DCT系数直方图的变化、DCT系数0游程的增加及对图像空域像素8×8块边界不连续性的变化等方面提取了用于检测F5算法的35个特征。最后利用SVM进行分类识别。本算法结合了专用隐写分析算法特征的有效性及通用隐写分析算法的多特征识别技术,对F5信息伪装的检测尤其是对低嵌入率图像的检测有很大的提高。实验仿真部分通过多种实验条件下的实验结果说明该算法具有很好的检测效果及很强的实用性。

Steganalysis for F5 Based on Multi-Characters

This paper analyzed F5 steganograghy,and gives 35 characters from DCT hists,zeros run-length,and blockiness measures which calculated from the decompressed JPEG images to detect F5 steganograghy.We use Support vector machines(SVM) to determine whether the image is a stego-image or not.The algorithm makes use of many effective characters.It has the merits of both specific and blind steganalysis.The experimental results from many conditions show that the algorithm is practical and works well,especially for detecting low embedding rate images.