An Integrated & Robust Image Forgery Detection Application using Neural Network
Nowadays it’s becoming easier and easier to tamper and edit an digital image due to the capabilities, availability of actively developing image manipulation software. Such easily available software enables anyone to edit and alter images and videos. Hence it is a case of necessity to verify and validate the authenticity of an image. During the process of creating a forged image a forger can manipulate the image using various image manipulation techniques like copy-move, object splicing, object removal and many more. Similar to these techniques there are various counter-techniques in the field of image forensics to detect each of these forgeries but with recent development in the neural networks it has become possible to integrate them into a singular solution. This paper proposes a highly robust forgery localization architecture which uses resampling features and deep learning. Furthermore, a large image splicing dataset is introduced to guide the training process. The proposed method is capable of localizing image manipulations at pixel level with high precision, which is demonstrated through rigorous experimentation on three diverse datasets.