Unexpected Artefacts in PRNU-Based Camera Identification: A "Dresden Image Database" Case-Study

TitleUnexpected Artefacts in PRNU-Based Camera Identification: A "Dresden Image Database" Case-Study
Publication TypeConference Paper
Year of Publication2012
AuthorsGloe, T., Pfennig S., & Kirchner M.
Page(s)109-114
Other Numbers3368
Abstract

Although sensor noise is generally known as a very reliablemeans to uniquely identify digital cameras, care has to betaken with respect to camera model characteristics thatmay cause false accusations. While earlier reports focusedon so-called linear patterns with a regular grid structure,also distortions due to geometric corrections of radial lensdistortion have recently gained interest. Here, we reportobservations from a case study with the `Dresden ImageDatabase' that revealed further artefacts. We found diagonalline artefacts in Nikon CoolPix S710 sensor noise, as wellas non-trivial dependencies between sensor noise, exposuretime (FujiFilm J50) and focal length (Casio EX-Z150). Atslower shutter speeds, original J50 images exhibit a slighthorizontal shift, whereas EX-Z150 images exhibit irregulargeometric distortions, which depend on the focal lengthand which become visible in the p-map of state-of-the-artresampling detectors. The observed artefacts may providevaluable clues for camera model identication, but also callfor particular attention when creating reference noise patternsfor applications that require low false negative rates.

Acknowledgment

This work was partially funded by the Deutscher Akademischer Austausch Dienst (DAAD) through a postdoctoral fellowship.

URLhttp://www.icsi.berkeley.edu/pubs/vision/ICSI_artefacts12.pdf
Bibliographic Notes

Proceedings of the ACM Multimedia and Security Workshop (MM&Sec 2012), pp. 109-114, Coventry, United Kingdom

Abbreviated Authors

T. Gloe, S. Pfennig, and M. Kirchner

ICSI Research Group

Vision

ICSI Publication Type

Article in conference proceedings