拷贝—变换—移动篡改是一种操作简单但非常有效的数字图像篡改方法,现有检测框架无法检测。同时，在处理高分辨率篡改图像时，基于穷举搜索的现有算法框架会有计算量的困难。针对高分辨率彩色图像拷贝—变换—移动篡改提出了基于图像融合和尺度不变特征变换(SIFT)特征点匹配的检测算法。首先利用基于加权多尺度基本形的图像融合方法将降采样后的彩色图像信息融合至单幅灰度图像中，然后设计快速匹配方法得到融合图像中匹配的SIFT特征点作为种子点，最后根据图像处理规则和SIFT特征点的尺度和方向信息制定合适的生长策略逐步生长出被篡改区域。对手动制作的篡改图片和可疑新闻图片的实验结果表明，该算法对常用的润色操作和亮度调整以及JPEG有损压缩有较强的稳健性。

Copy-transform-move forgery is a simple and powerful digital image forgery technique in which a region is copied and transformed and then pasted in another position on the same image. Copy-transform-move forgery defeats all existing copy-move forgery detection methods. The high resolution of forged images brings obstacles to existing exhaustive search based detection methods. In this paper, copy-transform-move forgery of high resolution color images is studied and a detection method based on image fusion and scale-invariant feature transform(SIFT) points is proposed. Firstly, the high-resolution image is down-sampled. Secondly, the proposed method fuses the multi-channel down-sampled image information into a single gray-level image. Then a fast matching method is designed to find matching SIFT points in the fused images. The forged regions grow up stepwise based on an appropriate region growing strategy with matching SIFT points as “seed-points”. The method is tested using man-made high-resolution forgery images and doubtful news pictures. Experiment results show that the method is robust to common retouch tools, brightness adjustments and lossy compression.