Statistical Error Estimation for an Objective Measure of Similarity to a Latent Image

As submitted by the proposer: This research project introduces an objective measure of similarity between a latent image and an exemplar, requiring no minutiae markup. The proposal provides evidence that the objective measure is shown to be very accurate when applied to the latent and True-Mate images from the NIST SD27 latent data set. This research project will use the SD27 set to demonstrate that current technology can compute a non-minutiae-based measurement of similarity to a latent together with an associated error statement concerning that measurement. This research project will use a novel latent image examination tool to create a non-minutiae based objective measure of the similarity between an exemplar image and a latent image. Further, through computing this similarity measure for competitive known non-mate exemplars, a statistical model is created that can be used to assess the rarity of any case exemplar. The rarity statement will be expressed in the context of an atypicality index relative to measured similarity to the latent image for known non-mate exemplar images. This research will provide valuable products for forensic science researchers who study the interpretation of evidence and inference concerning the source of a trace of unknown origin. The project will produce and evaluate statistical models based on each of the good, bad, and ugly latent images from the SD27 set. The latent image examination tool used in this project finds the most appropriate location to overlay the latent image onto any exemplar image, and then produces a distortion eliminating WARP of the latent image to the exemplar image. These operations are performed on automatically image processed latent and exemplar images. Note: This project contains a research and/or development component, as defined in applicable law, and complies with Part 200 Uniform Requirements - 2 CFR 200.210(a)(14). ca/ncf