Printer and Sensor Forensics

Principal Investigator: Jan Allebach; Ed Delp

This research addresses the need for a means to assure the authenticity of digital media consisting of image content. The work investigates both intrinsic signatures that are an inherent characteristic of the imaging device and extrinsic signatures that can be introduced by the manufacturer with the possibility including additional user-controlled information. The intrinsic signature represents artifacts that are due to optical, electrical, or mechanical limitations of the imaging device. The extrinsic signature is generated by modulating parameters that control the intrinsic signature of the device. The same algorithms that detect the intrinsic signature will form the basis for detecting and decoding the extrinsic signature.

This research will result in a new understanding of the relation between imaging devices and artifacts produced by those devices. It will lead to new knowledge regarding image analysis for feature extraction and the design of classifiers based on those features. In calculating error control codes and channel capacities for extrinsic signatures, it will extend the application of classical communications theory to a new domain. This work will be of direct benefit to society by providing law enforcement and government agents new tools for combating counterfeiting, forgery, and other criminal and terrorist activities.

Keywords: forensics, device, signatures, counterfeiting, forgery