摘要
针对当前基于JPEG图像的典型隐写算法,基于离散余弦变换(DCT)域分块内及分块间相邻系数之间的相关性进行分析,通过将相邻三个系数的中间位置的系数作为条件,统计两边系数的条件分布概率矩阵,将提取的所得矩阵作为隐写检测敏感特征,提出了一种基于DCT系数双边转移概率分布的JPEG图像隐写检测算法。实验结果表明,在不同嵌入率下,该算法的检测性能均优于已有检测算法。
For the typical steganographic algorithms in JPEG images,this paper firstly analyzed the correlation between neighboring coefficients of intra-and inter-block in Discrete Cosiine Transform(DCT) domain,and then extracted the conditional distribution probability matrix of the bilateral coefficients as the sensitive steganalytic features by taking the middle coefficient of three neighboring coefficients as the condition.At last,this paper proposed a JPEG image steganalytic algorithm on a basis of bilateral transition probability distribution of DCT coefficients.The experimental results show that,for different embedding ratios,the algorithm proposed in this paper outperforms the existing algorithms.
出处
《计算机应用》
CSCD
北大核心
2013年第4期1074-1076,1107,共4页
journal of Computer Applications
基金
河南省重点攻关项目(082102210067
112102210408)
关键词
隐写
JPEG图像
系数相关性
双边转移概率
隐写检测
steganography
JPEG image
coefficients correlation
bilateral transition probability
steganalysis