Image Forensics Evaluation Framework and Tool Testing

Abstract

The phrase ‘seeing is believing’ has been validated to the point where any proposition to the contrary sounds bizarre. The boom of the digital camera, photography, and social media has drastically changed how humans live their day-to-day, but this normalisation has been accompanied by malicious agents finding new ways to forge and tamper with images. Primarily, the motivation is unfair or unlawful monetary gain. Disinformation in the photographic media realm is an urgent threat. There are so many image editing tools available today that it is almost impossible to differentiate between a photo-realistic and an original image. The tools available for image forensics require a standard framework against which they can be evaluated. Such a standard framework can aid in evaluating the suitability of an image forensics tool for use in a criminal investigation, commercial operation, or for academic research. This research work proposes an evaluation framework for image forensics tools. The proposed framework is based on the conformance methodology of testing which employs test assertions and test cases. It is then tested by evaluating four image forensics tools namely FotoForensics, Ghiro, Imago Forensics, and Exif Reader. The framework provides a comparative insight into the tools based on test results. The evaluation of the image forensics tools revealed that FotoForensics provides a lot of optional features efficiently in addition to core features. The test results of Ghiro conformed to its usability features while Imago Forensics and Exif Reader lacked in providing a majority of optional features. This comparison can provide the information necessary for users to make intelligent choices about tools and it can help vendors shortlist areas of improvement in their tools.