Forensic biometric authentication.

With the advent of video-based surveillance systems, an increasing number of cases are presented to the court in which video footage or still image material presents possible evidence of the suspect being present at the crime scene. In these cases, identification ("who is het perpetrator?") or authentication ("Is the suspect the perpetrator?") of the individual can be based on comparison of morphological characteristics that can be extracted from the video or photographic material, such as body length or more detailed characteristics such as the face or even ears and footprints. In order to increase the value of photographic evidence in forensic cases this project first aims at maximizing the quality of the information present in forensic image data using advanced image enhancement and photgrammetric procedures. The second essential component of the proposed research is the use of 3D models and their inter- and intra-subject variations in the comparative photographic analysis circumventing the deficiencies of current 2D comparative image analysis. Third, the project also aims at assessing the evidential value of the available image material by investigating ways to assess Likelihood Ratios, expressing the strength of the evidence in supporting the hypothesis, giving all sources of uncertainty. While the project will focus on testing the implemented procedures mainly for height estimation, facial image comparison and bite mark analysis, the technology proposed in this project (image enhancement, photgrammetry, 3D (deformable model based) matching) can be adapted to other forensic biometric identifiers such as still image or video images of ears, latent ear marks, latent fingerprints, foot prints or even non-biometric identifiers such as shoe marks, tool marks and tire impressions. The proposed procedures have many applications outside the forensic context (medical, computer graphics, entertainment), as well, demonstrating the generic character of the methodology involved.

Keywords:Biometrics, identification, Authentication, Image enhancement, Model-based image analysis

Disciplines:Computer hardware, Computer theory, Scientific computing, Other computer engineering, information technology and mathematical engineering