U.S. flag

An official website of the United States government, Department of Justice.

Evaluation of an Emerging Automated Searching Technology to Improve the Efficiency and Reliability of Latent Print Comparisons

Award Information

Award #
2016-DN-BX-K004
Funding Category
Competitive
Location
Congressional District
Status
Closed
Funding First Awarded
2016
Total funding (to date)
$216,225

Description of original award (Fiscal Year 2016, $216,225)

As submitted by the applicant: In complex latent print comparisons, accurately matching latent prints that contain high levels of distortion or limited detail can be challenging to examiners. Research has shown that complex latent prints can lead to a decrease in the reproducibility and repeatability of examiner conclusions. Further, cases that contain a large number of comparisons have been shown to have an effect on examiner performance, where examiner fatigue may play a role. Siometrics LatentSleuth technology utilizes a ridge-skeleton matching algorithm to automate searching, which incorporates all usable information in a latent print as compared to minutia-point-only based approaches. The use of ridge geometry to create an overlay onto a reference print utilizes available information in a latent print even in the absence of traditional minutia. This allows for the matching of more challenging latent prints, including those that are stretched, warped, discontinuous, of an unknown orientation, or have few minutia points. Incorporating this technology into the current workflow could both reduce the overall search time and the total number of comparisons performed by an examiner by prioritizing the most likely candidates first.
The goals of this research project are to 1) determine the accuracy and reliability of the LatentSleuth technology, and 2) determine if integrating the LatentSleuth into current comparison workflow for complex comparisons improves a) efficiency and b) reproducibility as compared to existing methods. The software would initially be validated using ground truth reference samples. Following validation the LatentSleuth technology would be used on a dataset from casework that contains latent prints of a defined level of complexity or latent prints requiring a large number of comparisons. The results will be evaluated against traditional search and comparison techniques comparing efficiency, reproducibility and, potentially, accuracy. The dataset would contain unexamined latent prints from crime scenes and evidence collected in Virginia. Overall, our research will evaluate the benefits and limitations of incorporating the LatentSleuth technology into the comparison methodology for complex latent print comparisons.
Note: This project contains a research and/or development component, as defined in applicable law.
ca/cf

Date Created: September 12, 2016