NCJ Number
228231
Journal
Journal of Forensic Identification Volume: 59 Issue: 5 Dated: September/October 2009 Pages: 475-501
Date Published
October 2009
Length
27 pages
Annotation
This paper evaluated a statistical model based on sensor noise in the identification of digital cameras.
Abstract
The result of the evaluation on digital camera identification was shown to be promising for both charge-coupled device (CCD) and complementary metal-oxide-semiconductor (CMOS) cameras with relatively low noise levels. Uncompressed images and scaled images provided good results; however, the more compression that was applied to the images, the more the results degraded. Removing the row and column noise from the reference noise improved the result. Future tests are recommended involving more cameras of different models, quality, and types. In the forensic field, there is the possibility of being able to identify the camera that shot a specific image. It is claimed that the noise pattern is useful when determining whether a specific camera took a particular image. A statistical model was made by extracting the noise pattern associated with a camera and then comparing it to images from the same and other cameras. This paper evaluated this method based on the unique sensor noise in digital cameras. Figures, table, and references