When an attacker wants to falsify an image, in most of cases she/he willperform a JPEG recompression. Different techniques have been developed based ondiverse theoretical assumptions but very effective solutions have not beendeveloped yet. Recently, machine learning based approaches have been started toappear in the field of image forensics to solve diverse tasks such asacquisition source identification and forgery detection. In this last case, theaim ahead would be to get a trained neural network able, given a to-be-checkedimage, to reliably localize the forged areas. With this in mind, our paperproposes a step forward in this direction by analyzing how a single or doubleJPEG compression can be revealed and localized using convolutional neuralnetworks (CNNs). Different kinds of input to the CNN have been taken intoconsideration, and various experiments have been carried out trying also toevidence potential issues to be further investigated.
展开▼
