Forensic Imaging for Art Diagnostics. What Evidence Should We Trust?

Diagnostics digital images are often used to assess artworks. However, as all digital images they are also concerned by the issue of integrity. Computer vision techniques can be employed to obtain physical evidence of possible tampering. In this paper we explore the possibility to apply state of the art forensic algorithms to typical painting diagnostic images, taking into consideration real case studies. State of the art algorithms have been applied to genuine and modified diagnostic images to detect if, and how, forgeries of such images could be automatically detected and documented. To the best of our knowledge, this is the first time that such investigation is made. Results of the aforementioned tests prove that automatic assessment of the integrity of diagnostic images is challenging and that there are no reliable solutions currently available.