Deep Learning-Based Approach for Detecting Copy-Move Forgery in Digital Images
Naseeb Dar, Rani Atsya, Anuj Kumar · 2024
Copying and pasting a portion of the same image to hide another portion is the most common method of fraudulent image enhancement. Cloning a restricted portion of an image and pasting it once or more inside a comparable picture is known as copy-move image forgery (CMF). The procedure aims to highlight or hide a certain aspect in the finished picture, usually a person or an object. The harm brought about by the horrible deed frequently takes unexpected forms. For the automated identification of CMF, the paper proposes a Generative Adversarial Network (GAN) and Convolutional Neural Network (CNN) approach. The CASIA 2.0 dataset’s, which includes both real and fake photos, was used in the study. The proposed model is simple to use, and its application shows that it is faster than previous approaches that are now regarded as state of the art. The results show that, when compared to the other currently available methods, the recommended approach achieves the highest level of accuracy ($\mathbf{90.56 \%}$).