Demographic Effects in Latent Fingerprint Matching and their Relation to Image Quality

• NCJ Number: 305847
• Author(s): Emanuela Marasco; Mengling He; Larry Tang ; Yuanting Tao
• Date Published: 2022
• Length: 10 pages
• Grant Number(s): 2019-DU-BX-4011; 2019-DU-BX-0011
• Agencies: NIJ-Sponsored
• Sponsoring Agency: National Institute of Justice (NIJ)
• Publication Format: Article
• Publication Type: Research (Applied/Empirical)
• Country: United States of America
• Note: Presented at the 7th International Conference on Machine Learning Technologies (ICMLT), March 2022.

Annotation

In their previous work, the authors conducted an ROC regression analysis related to the demographic effects on latent fingerprint matching without considering the impact of image quality, and in this project, they extract quality measures from latent prints and consider them in the predictive model as an additional covariate to the demographics.

Abstract

Experiments were conducted using the FBI WVU BioCop 2008 database that contains 469 right-thumb and 219 right-index latent fingerprint images with associated demographics. Quality is estimated using the latent fingerprint image quality (LFIQ) algorithm. The findings show that the proposed covariate-adjusted ROC curve conditioned on image quality and demographics is a more informative assessment scheme than an evaluation without quality. (Publisher abstract provided)