Detection of Copy-Move Forgery Using DoGCode

Abstract

Copy-move forgery is performed as a part of an image is copied and moved to another region within the same image. For the detection of this type of forgery, usually, feature vectors of the image are created by dividing the image into sub-blocks. Then the feature vector of the region is compared with other regions in the image to reach a conclusion if they are matched. In this study, to determine the copy-move forgery, DoG Coding technique which is more resistant to changes in image brightness is used. Coded new image is formed with the help of this technique often preferred in the palmprint recognition systems and features are determined by dividing the image into sub-blocks. Block based region growing segmentation method have been developed to determine the size of the copied region. Proposed this new approach has been successfully applied on images in the CoMoFod dataset prepared by different scenarios.