基于分片变换的图像Copy-Move篡改检测方法  

Copy-Move Forgery Detection Using Slicing Transform

作　　者：谢伟[1,2] 万晓霞[1,2] 严文婧 叶松涛[3] XIE Wei WAN Xiaoxia YAN Wenjing YE Songtao(School of Printing and Packaging, Wuhan University, Wuhan 430072, China China Research Center for Photographic Culture Industry, Wuhan University, Wuhan 430072, China The College of Information Engineering, Xiangtan University, Xiangtan 411105, China)

机构地区：[1]武汉大学印刷与包装系,湖北武汉430072 [2]武汉大学中国摄影文化产业研究中心,湖北武汉430072 [3]湘潭大学信息工程学院,湖南湘潭411105

出　　处：《武汉大学学报（信息科学版）》2017年第9期1337-1342,共6页Geomatics and Information Science of Wuhan University

基　　金：国家重点基础研究发展规划(973计划)(2012CB725302);国家自然科学基金(61275172)~~

摘　　要：针对图像区域复制(copy-move,C-M)篡改检测方法通常面临的特征向量维度高、计算量大等问题,提出了一种基于分片变换(slicing transform,SLT)的图像C-M篡改检测方法。通过对目标图像进行SLT变换并对图像分片进行分组合并,提取每组图像分片的局部分片密度特征对图像块进行C-M篡改检测。实验结果表明,该方法提出的局部分片密度特征向量能够较好地表征图像特征,较典型的基于分块的篡改检测方法特征维度低,具有较低的时间复杂度和较高的检测率,并且对图像篡改区域的旋转攻击和缩放攻击亦具有较好的鲁棒性。Abstract： There are usually some issues such as high-dimension and high computational complexity that need to be solved in digital image copy-move forgery detection. In this paper, we propose a new method for digital image copy-move tampering detection based on slicing transform （SLT）. The original image is divided via Slicing Transform, then image slices is grouped before overlapping blocking. The local slice density feature vector of each group slicing block is extracted for copy-move forgery de tection. Experimental results demonstrate that the extracted feature vector which has lower dimension compared with other methods representing the image block. The proposed method can detect copy - move forgery regions faster and more accurately, and it is robust to post-processing such as rotation and scaling. Key words： SLT; copy-move; forgery detection; digital image forensics

关 键 词：SLT C—M 篡改检测 图像取证 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]