基于择多准则的新隐写分析方法

New steganalysis approach based on majority theory

为了达到更为可靠的检测,综合利用现有的检测算法,根据择多准则提出了一个择多分类算法的简化模型,得到了择多分类算法的理论错误分类概率的表达式,错误分类概率的上界和分类效果与检测算法数目的关系。应用该算法到图像空域LSB(least significant bit)隐写的分类问题,得到的实验结果表明该算法在一定程度上改善了分类效果.

To detect more reliably synthesizing , this paper gave a simplified frame with the majority theory. It got the expression and the upper bound of the new method＇ s theoretic wrong classifying probability along with the relationship between the classifying effect and the number of detecting methods, Applying the frame into ESB steganalysis in spatial domain, the experiment＇ s result shows that the new approach improves the classifying effect in a sense.