Open set source camera attribution and device linking
Pattern Recognition Letters
Hi-index | 0.00 |
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.