基于混合灰度序模式的图像复制-粘贴篡改盲鉴别算法  

Copy-move forgery detection based on mixed intensity order pattern

作　　者：朱叶[1,2] 申铉京[1,2] 陈海鹏[1,2] 

机构地区：[1]吉林大学计算机科学与技术学院,长春130012 [2]吉林大学符号计算与知识工程教育部重点实验室,长春130012

出　　处：《吉林大学学报（工学版）》2017年第4期1280-1285,共6页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金青年基金项目(61305046;61602203);吉林省自然科学基金项目(20140101193JC;20150101055JC)

摘　　要：针对目前复制-粘贴篡改盲检测算法对光照变换操作鲁棒性较差的问题,提出了一种基于混合灰度序模式(Mixed Intensity Order Pattern,MIOP)的复制-粘贴篡改盲鉴别算法。首先,对待检测图像提取高斯差分区域(Difference of Gaussians,DOG)。其次,利用MIOP特征描述区域。最后,匹配特征并利用RANSAC(RANdom SAmple Consensus)去除误匹配,确定图像的复制-粘贴篡改区域。实验结果表明:本文算法不仅对几何变换和光照操作的检测率较高,且对高斯模糊、噪声和JPEG重压缩等后处理操作鲁棒性较好。In order to solve the low robustness of copy-move forgery detection with illumination change,a novel method based on Mixed Intensity Order Pattern(MIOP)was proposed.First,the Difference of Gaussian(DOP)regions of the image to be detected are extracted.Second,the regions are described using MIOP.Third,the features are matched and the false matching is removed by RANdom SAmple Consensus(RANSAC).Finally,the copy-move forgery regions are detected.Experimental results show that the proposed method not only has high accuracy on geometric transformation and illumination change,but also has high robustness on post-processing operations,such as Gaussian blur,noise and JPEG recompression.

关 键 词：计算机应用 盲鉴别 图像复制-粘贴篡改 光照变换 混合灰度序模式 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]