Score-based likelihood ratios are a statistical method for quantifying the weight of evidence and have been used in many areas of forensics, including camera device identification1,2,3. Small sensor imperfections caused during manufacturing, called photo response non-uniformity4, leave identifying features, called a camera fingerprint, in the images that a camera takes. The sample correlation measures the similarity (or dissimilarity) between the camera fingerprint from the person of interest’s camera and the camera fingerprint in the questioned image. On its own, it is difficult to know how to interpret this score. Is a score of 0.25 evidence that the questioned image originated from the person of interest’s camera? What about a score of 0.5? To make sense of the score, it is compared with two different reference sets of scores: matching and non-matching. Matching scores are sample correlations between two fingerprints known to come from the person of interest’s camera. Non-matching scores are sample correlations between two fingerprints known to come from two different cameras. An alternative set of cameras that does not include the person of interest’s camera is used to build the set of non-matching scores. It turns out, that researchers have not agreed upon a best method for constructing the alternative population for score-based likelihood ratios5,6. Recently, researchers calculated score-based likelihood ratios for camera device identification using 48 cameras representing 26 models7. This present research explores whether the rates of misleading evidence can be decreased by restricting the alternative device population to cameras of the same brand as the person of interest’s camera.
Score-Based Likelihood Ratios for Camera Device Identification Using Cameras of the Same Brand for the Alternative Device Population

Conference/Workshop:
Annual Scientific Conference of the American Academy of Forensic Sciences (AAFS)
Annual Scientific Conference of the American Academy of Forensic Sciences (AAFS)
Published: 2022
Primary Author: Stephanie Reinders
Secondary Authors: Danica Ommen, Abby Martin, Alicia Carriquiry
Type: Presentation Slides
Research Area: Digital
Related Resources
Likelihood ratios for categorical count data with applications in digital forensics
We consider the forensic context in which the goal is to assess whether two sets of observed data came from the same source or from different sources. In particular, we…
CSAFE Project Update & ASCLD FRC Collaboration
This presentation highlighted CSAFE’s collaboration with the ASCLD FRC Collaboration Hub.
Forensic Analysis on Android Social Networking Applications
This presentation is from the 75th Anniversary Conference of the American Academy of Forensic Sciences, Orlando, Florida, February 13-18, 2023. Posted with permission of CSAFE.
Source Camera Identification on Multi-Camera Phones
Camera identification addresses the scenario where an investigator has a questioned digital image from an unknown camera. The investigator wants to know whether the questioned image was taken by a…