基于新型校准度量的F5隐写分析算法  

Steganalysis algorithm of F5 based on the new calibration technology

作　　者：张焱[1] 张敏情[1] 雷雨[1] 吴旭光[1] 

机构地区：[1]武警工程大学电子技术系网络与信息安全武警部队重点实验室,陕西西安710086

出　　处：《华中科技大学学报（自然科学版）》2013年第S2期227-231,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61379152);陕西省自然科学基金资助项目(2012JM8014);武警工程大学课题研究基金资助项目(2013S070)

摘　　要：通过对JPEG隐写分析中校准技术的研究,提出一种新的校准设计原则,并利用M-S(Mumford-Shah)模型引入新的校准方法,提出一种针对F5的隐写检测算法.算法首先从待测图像中提取M-S估计图像作为第一幅校准图,然后利用剪切重压缩法从隐写图中得到第二幅校准图,利用两幅校准图提取新型校准表示下的分类特征集.实验表明:该算法具有较好的检测性能,在嵌入率为5%时仍取得较高的检测率.Through the research on calibration technology in JPEG(joint photographic experts group) steganalysis,a new design principle of calibration was proposed,then the M-S(Mumford-Shah)model was utilized to get a new method of calibration and a new steganalysis algorithm of F5was proposed.This method first picked up a estimate image of the carrier based M-S as the first calibration image,then got the second image from the crop and recompress method proposed by Fridrich,with the two images,the feature sets was picked up according to the new form of calibration.The results show that this method has better detection effect and can obtain a high detecting rate even when the embedding rate is 5%.

关 键 词：校准 设计原则 隐写分析 M-S模型 算法 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]