DIGITAL CAMERA IDENTIFICATION BY MODELING THE PROBABILITYDISTRIBUTION OF SENSOR PATTERN NOISE

Sensor pattern noise (SPN) as the fingerprint of imagingdevices, could be used as a reliable feature in digital sourceidentification. In this paper, we introduce a new methodwhich uses the probability model of SPN to identify thesource. To achieve this goal, after extracting the SPN ofsome images, they are digitized and then for each value, thedistribution of its neighbors is modeled separately (Value-Model). Finally by using the value-model, the quantizedSPN is mapped to the probability domain. The average ofprobability matrix of some images of same camera formscamera model. This SPN model causes noticeable increasesin the true detection rate of the source. To evaluate theefficiency of proposed approach, we do some benchmark onour hypothesis. The accuracy and performance of our modelcompared to similar works, proves high efficiency of theproposed theory.

Keywords:

digital forensics , inherent pattern , sensor pattern noise , source identification , modeling