Graduate School and Research Center In communication systems

Seminar: Digital image forensics: sensor pattern noise for image authentication

Wiem TAKTAK - MM department intership student

Multimedia Communications

Date: April 4, 2011

Location: Eurecom - salle EN05

Image authentication is a crucial problem in digital image forensics. Using sensor pattern noise as a camera biometric trait provides promising results in identifying the camera that captured the image. In our work, we adopted the camera source identification approach based on sensor pattern noise and we proposed an enhancement process before generating the camera reference pattern noise. The presence of this sensor fingerprint in a given image was then checked using a correlation detector. Furthermore, given the camera source, a forged region of a given image can be detected using our proposed tampering detection method. Experiment show that this approach significantly outperforms state-of-the-art techniques and that it successfully identifies different images from the 'Dresden Image Database' which is available for forensic benchmarking.

Digital image forensics: sensor pattern noise for image authentication

Search