Differential Imaging Forensics: A Feasibility Study
share
We motivate and develop a new line of digital forensics. In the meanwhile, we propose a novel approach to photographer identification, a rarely explored authorship attribution problem. We report a proof-of-concept study, which shows the feasibility of our method. Our contributions include a new forensic method for photographer de-anonymization and revealing a novel privacy threat which had been ignored before. The success of our creation builds on top of a new optical side-channel which we have discovered, as well as on how to exploit it effectively. We also make the first attempt to bridge side channels and inverse problems, two fields that appear to be completely isolated from each other but have deep connections.
READ FULL TEXT