Information Content of Friction Ridge Impressions as Revealed by Human Experts

• NCJ Number: 244567
• Author(s): Thomas Busey; Chen Yu
• Date Published: January 2014
• Length: 65 pages
• Grant Number(s): 2009-DN-BX-K226
• Agencies: NIJ-Sponsored
• Publication Type: Research (Applied/Empirical); Report (Study/Research); Report (Grant Sponsored)

Annotation

This project constructed a quantitative representation of the information content in fingerprints.

Abstract

This representation has two benefits. First, it reveals those areas of a fingerprint that experts consider most diagnostic, which can assist trainees or jurors in deciding the value of different regions of the print. Second, it can guide experts on how diagnostic a region may be if it differs from their expectations. These benefits are particularly important when a suspect is identified through a database search that uses a system such as the Automated Fingerprint Information System (AFIS). In such searches, high similarity between the print collected in the investigation and a candidate from AFIS should be viewed with skepticism, because AFIS is designed to return all prints that look similar to the ones submitted for comparison. Thus, a quantitative representation of the information content in fingerprints would assist in identifying which regions of the prints are most diagnostic, given the statistics of the entire database. The challenge is to discover a feature representation that will allow statistical or quantitative analyses. This technical report on the project's work describes the nature of the eye-tracking data collected from experts and novices, how this is used to train a set of intermediate-level descriptors shown as "basis functions," and how the activations of these basis functions can be used to provide a quantitative description of the information contained in friction ridge impressions. A major contribution of this project is the development of robust eye-tracking methods that allow the collection of eye gaze data in the field as examiners conduct tasks that are similar to casework. The report discusses the development of the tools that allow this data collection. 40 figures and 13 references