Identifying the authenticity and integrity of digital images becomes increasingly important in digital forensics. In this paper, we propose an effective method for detecting copymove forgery. This method is implemented by first extracting SV features, which are invariant to algebraic, geometric changes, and some disturbances. Due to similar texture characteristic between copied and pasted regions, each SV feature vector is represented as a query and is then matched to its nearest neighbors in image. Experiments are provided to demonstrate the efficiency of presented method on different forgeries and evaluate its robustness and sensitivity to tampered images with some post image processing.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below