This document reports on a comprehensive research study of post-mortem iris recognition, both in terms of application of Artificial Intelligence methods as well as involvement of human subject analyzing the iris samples.
The author of this report addresses the main question of how to establish a possibility for performing post-mortem, forensic iris automatic recognition, and how to assist human examiners in their efforts. The research project aimed at delivering a complete methodology and software that can enhance unidentified decedent systems with a capability for comparisons of perimortem and postmortem iris images. During the research study, the author applied various computer vision techniques for performing post-mortem iris image processing, including segmentation, encoding, and matching. Among the methods developed, the author established the benefits of purely deep learning-based solutions as well as those that incorporate human intelligence in the method design for matching PM images with AM or perimortem images. Additionally, the established method for examining fingerprints was crucial for designing the research methodology of the human-machine paring for human examination of forensic iris samples. Results from this research bring the possibility of iris recognition as an identification method, and the expansion of iris image datasets, one step closer, by offering a software-based system for automatic post-mortem iris recognition, methodology that supports human examiners, and a new dataset of peri-mortem and post-mortem iris samples collected from 269 subjects. The designed software was delivered to the Dutchess County Medical Examiner’s office as well as the National Archive of Criminal Justice Data repository, and may serve as an element of forensic toolkit.
Downloads
Related Datasets
Similar Publications
- Identification of Blunt Force Traumatic Fractures in Burned Bone
- Criticality of Spray Solvent Choice on the Performance of Next Generation, Spray-Based Ambient Mass Spectrometric Ionization Sources: A Case Study Based on Synthetic Cannabinoid Forensic Evidence
- ILIAD: A Suite of Automated Snakemake Workflows for Processing Genomic Data for Downstream Applications