Detection of Region Duplication Forgery in Images under Affine Transforms

Region duplication is a common method to produce forgery images, where part of an image is copied and pasted somewhere else in the same image. In order to fit the scene better and leave no visible artifacts, the copied region may be processed by affine transforms before being pasted. Most of the existing methods cannot handle these transforms. This paper presents a method to detect the region-duplication forgery under affine transforms. The image is first filtered and divided into overlapping circular blocks. Then the normalized color histogram (NCH) is extracted as the block feature. Forgery detection is achieved by comparing the NCH features. A new filter is designed to process the initial detection results. The final detection map is obtained after morphological operations. Simulations demonstrate the efficiency of the method.