Digital Image Forensics
Overview
Digital image forensics asks whether images are authentic, how they were altered, where edits occurred, and sometimes which tool or source produced them. Methods examine noise, frequency, compression, and other pixel-level traces, often combining them with semantic features from deep networks or vision-language models; related work uses watermarks and steganalysis for authentication and hidden-data detection. Current systems increasingly aim to handle diverse images and tasks, including documents, video, and synthetic-media localization, with explainable evidence rather than only a real-or-fake score. A central challenge is generalizing across generators, editing pipelines, image domains, and post-processing, while avoiding misleading benchmark shortcuts and making calibrated claims about authenticity.
Latest papers
113 papers10 papers in the last four weeks, down 33% on the four weeks before. 0.1% of all new papers.