Copy-Move and Image Splicing Forgery Detection based on Convolution Neural Network

Abstract

Digital images plays a very significant role in fields like journalism, medical imaging, criminal and forensic investigations and many more. Because of the easily available photo editing tools and software, images can be manipulated easily, that can disturb the contents of the images. Due to this, authenticity of the image gets lost and these can be misused by any person. The techniques that are commonly used for creating forged images are copy-move, image splicing and image enhancement forgery. Many techniques were developed to detect forgery, but these techniques are not robust against the structural changes occurred due to forgery in the images. In this paper, Convolution Neural Network(CNN) based image forgery detection method is proposed. In this method, Patch Sampling and Modulus LBP will be used to pretrain the neural network for Feature Learning and Feature Extraction. Then finally these features will be fed to SVM classifier that will help to detect forged images.Evaluation of the proposed method is done based on the parameters like precision, recall and accuracy, which shows that the proposed method is robust and insensitive against different operations as well as there is the improvement in the accuracy of the proposed method as compared to existing method.