Integrity analysis and detection of digital forensic evidences

Abstract

Technology has improved people’s day to day activities like how we communicate and access information. These days people are equipped with a digital camera and mobile phone and they tend to record almost everything happening around them like capturing food they are having or capturing beautiful sceneries around them. Maintaining image integrity is not crucial in informal situations but it is very important to maintain for forensics scientists who are dealing with digital forensic evidence. Recently in our country, a new law has been passed which states that from now on digital proofs can be used in court as evidence. As we know, digital files can be modified; hence the authentication of each and every piece of digital evidence has to be verified manually by experts. There are some researches on this, but could not find any feasible publicly available datasets to work on tools that can detect tampered automatically. My target is to build a dataset consisting of copy-move and cut-paste image forgeries created from the original images; and build a system with CNN models that will detect and automatically exclude photographs that have obvious, and medium levels of modification which will ease the pressure on digital forensics scientists.