A review on blind detection for image steganography
A review on blind detection for image steganography
复制标题
DOI：
10.1016/j.sigpro.2008.03.016
复制
发表时间：
2008-09
期刊：
影响因子：
--
通讯作者：
Xiangyang Luo;Daoshun Wang;Ping Wang;Fenlin Liu
中科院分区：
文献类型：
--
作者：
Xiangyang Luo;Daoshun Wang;Ping Wang;Fenlin Liu
盲隐写分析技术是在隐写嵌入算法未知的情况下，检测数字媒体中是否存在嵌入的秘密信息。本文综述了数字图像的盲隐写分析方法。首先，描述了图像盲隐写分析的原理框架，包括图像预处理、特征提取、分类器选择与设计和分类四个部分。然后根据特征提取和分类器设计的发展将现有的盲检测方法分为两类。对于第一类，我们概述了六种典型特征提取方法的原理，简要描述了这些方法的特征提取算法，并利用Bhattacharyya距离对一些典型特征提取算法的性能进行了比较。第二类分类器设计的发展，我们对现有盲检测方法中使用的各种分类算法进行了综述，并详细介绍了基于多元回归分析、OC-SVM、ANN、CIS和超几何结构的几种分类器背后的算法。最后，对该领域有待解决的问题进行了讨论，并指出了未来值得研究的方向。
Blind steganalysis techniques detect the existence of secret messages embedded in digital media when the steganography embedding algorithm is unknown. This paper presents a survey of blind steganalysis methods for digital images. First, a principle framework is described for image blind steganalysis, which includes four parts: image pretreatment, feature extraction, classifier selection and design, and classification. We then classify the existing blind detection methods into two categories according to the development of feature extraction and classifier design. For the first category, we survey the principles of six kinds of typical feature extraction methods, describe briefly the algorithms of features extraction of these methods, and compare the performances of some typical feature extraction algorithms by employing the Bhattacharyya distance. For the second category, the development of classifier design, we make a survey on various classification algorithms used in existing blind detection methods, and detail the algorithms behind several classifiers based on multivariate regression analysis, OC-SVM, ANN, CIS and Hyper-geometric structure. Finally, some open problems in this field are discussed, and some interesting directions that may be worth researching in the future are indicated.