An analysis on attacker actions in fingerprint-copy attack in source camera identification

Multimedia forensics deals with the analysis of multimedia data to gather information on their origin and authenticity through the use of specific tools. At this point an important question arises: how much reliable are these algorithms? In this work we have considered the technique presented in [1] where it is shown how source camera identification can be attacked. In particular, the problem investigated concerns the situation when an adversary estimates the sensor fingerprint from a set of images belonging to the person he wants to frame and superimposes it onto an image acquired by a different camera to charge the innocent victim as the author of that photo. In [1], a countermeasure against such attack, named Triangle Test, is introduced. In this paper we have analyzed if a more sophisticated action of the attacker can invalidate such countermeasure. Experimental results are provided to prove how hacker's action could be improved.