Robust passive image authentication scheme based on serial companied approach

Due to the wide use of digital images in the virtual world and the growing attacks on them, passive image authentication schemes became an urgent and necessary need. The forgery attacks threaten the ownership of the digital image. Copy- move forgery detection is the main type of image forgery detection schemes. In this forgery, at least one region is copied and pasted at same image. In the proposed scheme, we proposed anew serial strategy through combining feature-based detector and block-based detector by using Region growing merging Segmentation method (RGMS).Multi scale Hessian-affine detector as primary detector, Polar cosine transform(PCT) as final detector and Simple linear iterative clustering segmentation(SLIC) with Factor Graph to construct RGMS. We designed serial random fitting model to include the true results and exclude the undesired results. Serial random supports single and multi-cloning regions. Our strategy provides facility to reveal and extract the iterative regions under different constraints. In addition, the proposed scheme is able to detect the slight and invisible cloning regions in the smooth digital image. This scheme is invariant to simple and complex attacks. Furthermore, it invariant to affine transformation likes skewness, distortion and perspective changes. Different datasets are used like GRIP, MICC-F600, MICC-F8.The experimental results confirmed that the proposed scheme has high percentage of True positive rates (99%), and low percentage of false positive rates (1.5%). The proposed scheme can be used as passive image forensic scheme in the cybercrimes and court rooms.