基于SNR和PSNR特征的净图定量描述方法的研究  

Research of the Cover Images Quantitative Describing Models Based on the SNR Feature and PSNR Feature

作　　者：周玉林[1] 毛家发[1] 熊鹏荣[1] 

机构地区：[1]上饶师范学院,江西上饶334001

出　　处：《上饶师范学院学报》2009年第3期81-84,93,共5页Journal of Shangrao Normal University

基　　金：江西省教育厅科技基金项目(Gjj08462);上海市教育发展基金会晨光计划(2008CGB21)

摘　　要：在隐写分析领域,绝大多数的分析方法都采用二类样本(干净图像与掩密图像)进行监督学习的。本文的研究紧紧地围绕干净图像样本,提出了基于SNR和PSNR特征的净图描述方法,通过这二种特征对净图进行量化描述,并在理论和实践中验证了用这二种特征来定量描述干净图像是合适的.当干净图像描述清楚了,干净图像与掩密图像自然而然地区分开了。这种新颖的净图定量描述方法,大大提高了隐写分析方法的适应性。In the domain of steganalysis, almost all of analysis method adopt binary sample to supervise study. （cover image and covert image）. The research of this paper revolves around the cover image sample,and bring up the feature diseription method of the signal to noise ratio（SNR） and the peak signal- to- noise rafio（PSNR） ,through this two kinds of feature to describe and quantification the cover image, and verified in theory and pracice ,it is suitable to use this two kind of features to discribe the cover image. Cover image and covert image naturally disinguishes when clean picture describe clear. This fresh ration method to net picture, has greatly increased adaptability to the seganalysis method.

关 键 词：隐写分析 干净图像 掩密图像 SNR特征 PSNR特征 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]