Back to search

West Virginia University

AFIS Based Likelihood Ratios for Latent Fingerprint Comparisons

Abstract

dc:description.abstract

Latent fingerprints are one of the most common pieces of evidence found on a crime scene and represent accidental or unintentional prints collected as part of a criminal investigation. They are caused when the friction ridge skin comes in contact with a surface, and thus requires the use of chemical processing to be visualized with the naked eye. The comparison and identification of fingerprints depends on various factors such as the substrate quality, surface, duration, environmental factors and examiner experience. These factors can result in reduced clarity or content, and can even cause distortions as compared to a fingerprint taken under controlled conditions. Since the release of the National Academy of Sciences (NAS) report in 2009, the field of fingerprint analysis has come under much scrutiny. Specifically, the need for more research into the determination of the accuracy and reliability of the identifications made by fingerprint examiners has been raised.;One such method used for the comparison of latent fingerprint to known prints is through an Automated Fingerprint Identification System (AFIS). The AFIS used in this research was the AFIX Tracker R where where variables were assessed: match score, match minutiae, match status, delta match score and marked minutiae, to determine which variable(s) was a better indicator of a true match. Bayesian networks were then constructed to compute the likelihood ratios to evaluate the dependency of the variables on one another,where the performance of the likelihood ratios in determining the identity of the unknown latent was assessed using Tippett and ECE plots. Receiver Operating Characteristic (ROC) curves and Bayesian networks were constructed to perform statistical analysis of the matches obtained while comparing a latent print to a ten-print card. A combination of Tippett and Empirical Cross Entropy (ECE) plots were used to assess the performance of the AFIX Tracker R in classifying unknown prints. It was observed that a match minutiae of 15 or higher resulted in a 100% true match result whereas for the non-matches,no more than 13 match minutiae were found. Moreover, the delta match scores difference between the matches and non-matches were notable (delta score of 0.1-153 for matches compared to a score of 0-0.1 for the non-matches). Overall, it was determined that approximately 87% of the time a randomly selected known match would have a higher number of match minutiae as compared to a non-match.

Degree

thesis:*
Name thesis:degree_name
MS
Level thesis:degree_level
Thesis
Discipline thesis:degree_discipline
Forensic and Investigative Science

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Kamath, Shreya Sateesh
Contributors dc:contributor
  • Keith B Morris
  • Afzel Noore
  • Jacqueline A Speir

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
West Virginia University
Base URL
researchrepository.wvu.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Kamath, Shreya Sateesh. AFIS Based Likelihood Ratios for Latent Fingerprint Comparisons. Thesis thesis, https://doi.org/10.33915/etd.5927