Publication:

Video camera identification using PRNU pattern noise as digital fingerprint

Nowadays the use of videos or images as clues in court for forensics is growing up, and camera identification is developing to make those clues as more reliable as possible. To achieve this a proper camera identification algorithm is needed. In this thesis one strategy is proposed to catch this objective. PRNU pattern noise is unique for each camera, so it will be considered as unique fingerprint for each camera. In this thesis one way of extraction of it is proposed, and after that the comparison of estimated fingerprint with the reference fingerprints of our given database is going to be implemented by using correlation mathematical operation, in order to find the highest coincidence. Several experiments are proposed in order to have a proper algorithm of camera identification. Mainly those experiments are based on detecting errors and anomalies.