摘要
为解决JPEG图像隐密分析算法的设计问题,在DCT域共生矩阵的基础上构建了JPEG图像共生特征空间,并将定量隐密分析算法看成多元共生特征与含密量之间的回归模型,提出了一种JPEG图像定量隐密分析算法设计方案.此方案以已知含密量的JPEG图像为研究对象,而无需分析对应的JPEG隐密机制,因此具有较强的通用性.实验结果表明,对安全性较高的MB1隐密算法,本文方案仍可设计出高检测能力的定量隐密分析方法.
A co-occurrence feature space is constructed, which can effectively reflect the changes of higher-order statistical characteristics caused by JPEG steganographic techniques. Supported by this feature space,a new scheme for designing steganalytic algorithms is presented based on multivariate regression model. With only JPEG stego images and no embedding details needed, this designing scheme is able to provide a good generality. Experimental results indicate that, our designing scheme can also produce remarkable steganalytic algorithm even for MB1 embedding mechanism, which has reached high performance of security.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2009年第6期1378-1381,共4页
Acta Electronica Sinica
基金
国家自然科学基金(No.60572111)
关键词
多元回归
JPEG
隐密分析
共生特征
MB1
multivariate regression
JPEG, steganalysis,co-occurrence feature,MB（model based）
