Research On Image Copy-move Forgery Technology Detection Based On Local Color Moments

With the advent of the digital age and the development of information technology,multimedia tools have been widely used,making it easy for people to edit images and videos.But if maliciously tampered images appears in some important scenes,such as magazines,medicine,law,etc.,it will bring serious consequences.Copy-move forgery detection technology is an important branch in the field of forensics,usually extracting features from an image and matching them with each other.The forgery regions have similarity after copy and move operations,so they can be located according to the similarity of the matching results.According to the study of some existing copy-move forgery detection technologies.Aiming at the problems of Patch Match algorithm that does not consider the semantic information between superpixels,insufficient keypoint extraction,poor recognition rate of feature descriptors,and unclosed tampering regions.Two algorithms are proposed in this paper.(1)An image copy-move forgery detection method based on local color moments and superpatch match is proposed.The SIFT keypoints of the image are extracted and matched.If the keypoints are similar,the Superpatch where the keypoints are located is also similar.Therefore,the Superpatch is constructed with the superpixel represented by the keypoints as the barycenter,and the Superpatch of the superpixel is matched by the Superpatch Match algorithm after combining with the barycenter registration,so as to obtain the detection results.The experimental results show that it has better detection performance than SIFT algorithm,PM algorithm and Yu et al algorithm,and has better robustness against multiple forgery attacks.(2)An image copy-move forgery detection algorithm based on local color moment and quaternion Hu moment is proposed.Firstly,the adaptive morphological reconstruction algorithm is used to divide the superpixels,and the density clustering algorithm is used to cluster the local color moments in the superpixels and realize the adaptive segmentation of the image region;Secondly,the SIFT keypoints are evenly distributed in an image by using the keypoint extraction method,and then the local Gaussian pyramid is constructed in the novel quaternion representation method of color image,and the Hu moment feature is extracted to enhance the feature description;After feature matching with 2NN,the copy move forgery regions are located by combining Delaunay triangle algorithm.Experimental results show that the accuracy of the algorithm is better than that of some existing algorithms,and it can effectively resist scaling and rotation attacks.