Digital Image Forensic Analysis using Hybrid Features

Copy Move Forgery (CMF) is an exceptional sort of image control technique in which any section of the input sample is copied into another piece of a similar image. Recently, several researchers proposed different techniques to deal with the authentication of images. However, still, there is a need for improvement both in terms of localization performance and time efficiency. In this paper, we presented a passive approach that is based on LDP with DCT for the CMF recognition. Initially, the image is transformed to YCbCr and then distributes into overlapping blocks. The features are extracted through LDP with DCT and after that similarity of the blocks is measured through Euclidean distance. The evaluation of our presented framework is performed on a challenging dataset namely CoMoFoD_Small_V2. The proposed method achieved a higher precision of 99% and a recall of 96% as compared to latest methods. Experiments show that our approach successfully detects the CMF parts and is more robust to post-processing operations.