2020/10/08 by Daniel Mas Montserrat, János Horváth, Montserrat, Daniel Mas +7
Computer Science · Engineering · #Advanced Image Processing Techniques #Computer Vision and Pattern Recognition (cs.CV) #Digital Media Forensic Detection #FOS: Computer and information sciences #Image Processing Techniques and Applications
paper · pdf · doi:10.48550/arxiv.2010.03758
openalex publication_date 2020/10/08 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Satellite imagery is becoming increasingly accessible due to the growing\nnumber of orbiting commercial satellites. Many applications make use of such\nimages: agricultural management, meteorological prediction, damage assessment\nfrom natural disasters, or cartography are some of the examples. Unfortunately,\nthese images can be easily tampered and modified with image manipulation tools\ndamaging downstream applications. Because the nature of the manipulation\napplied to the image is typically unknown, unsupervised methods that don't\nrequire prior knowledge of the tampering techniques used are preferred. In this\npaper, we use ensembles of generative autoregressive models to model the\ndistribution of the pixels of the image in order to detect potential\nmanipulations. We evaluate the performance of the presented approach obtaining\naccurate localization results compared to previously presented approaches.\n