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.%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.
展开▼