论文提出一种新的针对灰度图像的隐写分析算法.该方法对灰度图像进行水平和垂直差分运算,然后对得到的差分值计算共生矩阵,从共生矩阵中得到180维特征并使用支持向量机进行分类.该文使用UCID图像库,针对三种常用的隐写算法:Cox扩频算法,QIM算法以及LSB隐写算法进行试验,结果表明该法能有效地检测出上述三种隐写方法.“,”A new steganalysis scheme for gray images was proposed in this paper. A total of 180 dimensional feature vectors were derived from the cooccurrence matrix,which was calculated directly from horizon and vertical differential image. SVM was used to classify the 180 dimensional feature vectors. The experimental results for 3 kinds of popular steganographic schemes (Cox spread spectrum,QIM,LSB)have demonstrated that the proposed scheme outperforms the existing steganalysis techniques.