Copy Move and Splicing Image Forgery Detection using CNN
Devjani Mallick, Mantasha Shaikh, Anuja Gulhane, Tabassum Maktum
Abstract
The boom of digital images coupled with the development of approachable image manipulation software has made image tampering easier than ever. As a result, there is massive increase in number of forged or falsified images that represent incorrect or false information. Hence, the issue of image forgery has become a major concern and it must be addressed with appropriate solution. Throughout the years, various computer vision and deep learning solutions have emerged with a purpose to detect forgery in case of digital images. This paper presents a novel approach to detect copy move and splicing image forgery using a Convolutional Neural Network (CNN) with three different models i.e. ELA (Error Level Analysis), VGG16 and VGG19. The proposed method applies the pre-processing technique to obtain the images at a particular compression rate. These images are then utilized to train the model and further the images are classified as authentic or forged. The paper also presents the experimental results of the proposed method and performance evaluation in terms of accuracy.