基于矩特征的隐写分析方法
Steganalysis Based on Moment Features
-
摘要: 针对互联网上大量的隐写软件利用图像隐藏秘密消息,提出一种基于矩特征的隐写分析方法.这种隐写分析技术能有效地检测出图像是否含有秘密信息,通过多分辨率表示图像,在不失有效信息量的基础上减小了数据量的处理,提取2种新颖的、能反映图像隐写前后的统计特性变化的特征,即PDF矩特征和CF特征,对干净图像进行定量描述,并且设计了与特征相匹配的柔性化线性一类分类器,具有一定的稳健性和柔化特性,提高了检测性能.Abstract: As for a large number of steganographic programs hiding message in digital image on the internet,a steganalysis technique based on moment features was proposed in this paper.It could detect effectively whether or not secret messages were included in digital images by this technique.Multiresolution image representation could reduce the amount of data processing based on not missing effective information.Two new features were extracted,namely probability density functions(PDF) and characteristic functions(CF),which could reflect the change of statistical properties.And the clean image could be quantitatively decribed.A one classifier based a flexile linear was designed which had stable and soft feature.And the detection performance could be improved.