U.S. flag

An official website of the United States government, Department of Justice.

Neural Network System for Matching Dental Radiographs

NCJ Number
241123
Journal
Pattern Recognition Volume: 40 Issue: 1 Dated: 2007 Pages: 65-79
Date Published
2007
Length
15 pages
Annotation

This paper discusses a prototype automated dental identification system for use in postmortem identification.

Abstract

In 1997, an automated dental identification system (ADIS) in conjunction with a digital image repository (DIR) was proposed as a means for use in postmortem identification that would be similar in nature to that of the automated fingerprint identification system (AFIS). This paper presents a prototype ADIS that deals with the problems found in dental image matching. The paper begins with background on two topics closely related to ADIS: image matching and automated postmortem identification systems. This is followed by a discussion of the prototype ADIS. The prototype system carries out identification in two phases: the first phase is a fast, high-recall retrieval system, while the second phase is a high-precision matching system. Each of these phases is discussed in detail in the next section of the paper. The final section of the paper presents the results of training and testing experiments that were carried out represent and verify the micro-decisionmaking stage of the image comparison component of the prototype. The results indicate that the prototype ADIS showed matching accuracy of more than 80 percent. Plans for future research are discussed. Tables, figures, and references

Date Published: January 1, 2007