Publication

Abstract : The availability of the most advanced image correction tools along with the latest highly classy capturing devices has made the tampering of the images more easier. Copy move method of tampering happens when a particular section in the digital image is copy pasted on a new section on the same image for guarding or else hiding offensive areas. Digital forgery detection methods on images target the discovery of fake regions or the duplicated parts. Many pre- or post-operations are done on the images by persons who do the tampering. A novel method integrating the block-based and key-point-based methods is suggested in this paper. Initially, it adaptively segments the input host picture into non-overlapping and unequal sized slabs and features and thus mined from each block is harmonized with each further to discover labeled feature points that point out doubted fake regions approximately. In order to make it further precise, an extraction algorithm is proposed that substitutes the feature points by minor superpixels as slabs of feature which later fuses with the adjacent blocks which show like local color features to create that detected forgery sections. Lastly, morphological operations are performed on these fused areas to produce the spotted tampered areas. Experiments show that this method of detection achieves better outputs under several perplexing conditions compared with the latest CMFD algorithms.