Putting the PRNU Model in Reverse Gear: Findings with Synthetic Signals

The prevalent model for the photoresponse nonuniformity (PRNU) of digital cameras is (negatively) tested. The PRNU serves as a fingerprint of the underlying device that has proven its usefulness in image and video forensics. When such model is applied to derive analytical expressions characterizing the PRNU detection statistic, the predictions do not conform with reality. However, those expressions are thoroughly validated here through extensive experimentation using synthetic signals with parameters taken from real images. As a consequence, the current multiplicative PRNU model must be revised. This opens new venues for performance improvements in both the extraction and detection of device fingerprints.