Camera-based Image Forgery Localization using Convolutional Neural Networks

Camera fingerprints are precious tools for a number of image forensics tasks.\nA well-known example is the photo response non-uniformity (PRNU) noise pattern,\na powerful device fingerprint. Here, to address the image forgery localization\nproblem, we rely on noiseprint, a recently proposed CNN-based camera model\nfingerprint. The CNN is trained to minimize the distance between same-model\npatches, and maximize the distance otherwise. As a result, the noiseprint\naccounts for model-related artifacts just like the PRNU accounts for\ndevice-related non-uniformities. However, unlike the PRNU, it is only mildly\naffected by residuals of high-level scene content. The experiments show that\nthe proposed noiseprint-based forgery localization method improves over the\nPRNU-based reference.\n