Description of original award (Fiscal Year 2005, $431,255)
Quantitative approaches to fingerprint identification rely on different approaches derived from minutiae detection, orientation computations and other sources of information. These approaches extract out relevant features that can be matched across prints. The applicant proposes to integrate human expertise into a statistical model to perform a quantitative analysis of inked and latent prints. It will incorporate the visual search patterns of fingerprint examiners into machine learning algorithms to quantitatively analyze fingerprints.