Award Information
Description of original award (Fiscal Year 2018, $642,979)
The purpose of this project is to deliver a complete methodology and the software that enhance unidentified decedent systems with a capability for comparisons of antemortem and postmortem iris images. The main research problem concerns novel methods, both related to computer vision and human examination that allow for identification of deceased subjects based on their iris patterns. The final product will include (1) a methodology and its implementation as a software tool returning a ranked list of best matches between a postmortem sample and a gallery of antemortem and postmortem samples, with visualization of features that support computers judgment, and (2) a human examination methodology, which together with computer-added analysis of the returned candidate images, will deliver the final judgment about the identity of a decedent. Quarterly progress reports, the final report, and source codes with documentation of the software will be delivered. This project will be led by the University of Notre Dame (ND), which developed a software tool supporting human examination of postmortem iris images. Dutchess County Medical Examiners Office (DCMEO) will collect new postmortem and perimortem iris images and will verify the project's outcomes from the standpoint of medical examination requirements. Michigan State University will work with ND on new computer vision methods for postmortem iris processing. The team has a unique dataset of 6,100 postmortem iris images from 180 cadavers, with postmortem interval from a few hours up to 34 days, including 3 perimortem cases. In the first project phase, the existing data will be used to develop iris matching methodology that is specific to postmortem samples. The developed methods will be verified with data acquired from new 250 cadavers, including perimortem samples. In the second phase, the experiments at ND with human examiners analyzing perimortem and postmortem iris image pairs with different levels of difficulty will be conducted, and the methodology for human-expert-based analysis of the best-matching candidates will be developed. The ACE-V protocol used for comparison of fingerprints, will be adapted to human-based iris recognition. The project will generate a complete system, including methodology and supporting software, for forensic analysis of iris images.
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