摘要
为了对图像篡改中常用的复制-移动伪造进行检测,基于傅里叶-梅林变换的平移、旋转和缩放的不变性提出一种快速图像区域分割和匹配的高效篡改检测算法。不同于以往模板匹配方式中按照单像素点移动得到重叠块划分方法,该算法采用相邻图像块的图像区域分割方式来减少整个图像块的数量。通过相似性匹配检测,得到初步的复制图像区域,然后利用边缘处理的方法处理改善篡改区域,从而达到改进篡改检测算法的效率和准确性。最后通过实验验证了该算法的有效性。
To detect copy-move forgery commonly used in of image tampering,based on Fourier-Mellin Transform（FMT）,which is invariant to translation,rotation and scale type of manipulations,a quick image section segmentation and matching algorithm is presented.Unlike overlapped block segmentation with single pixel movement in other methods of template matching,this algorithm uses a different image section segmentation approach by neighbor image block to reduce the number of image blocks.After getting initial copy image sections through detecting the similar blocks,improve the tampering section by additional edge handling.Then,efficiency and accuracy for tamper detection algorithm can be achieved.The experiment results show that the proposed method can detect duplicated regions in the images very accurately.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第15期3530-3532,3536,共4页
Computer Engineering and Design
基金
浙江省科技厅重点科研计划基金项目(2008C21087)
浙江省自然科学基金项目(Y3090061)
