Image Forgery Detection Using Colour Moments

With the rapid development of powerful image editing software, digital image forgery has been a serious problem. One of the most commonly used forgery technique is the Copy-move forgery that copies a part of an image and pastes it on the other region in the same image. Some methods in the literature divide suspicious image into overlapped blocks and extract some features from them to judge the forgery. Similarity among the feature vectors gives a clue about the forgery. In this work, we used first three-color moments to extract feature vectors from the blocks. The method assumes that the color distribution of a block cannot be changed even if it is compressed or blurred. Color Moments has not been used to detect image forgery in the literature before. The proposed method has higher accuracy ratios compared to other works when the forged image is post processed using some operations.