Matching of Dental X-rays for Human Forensic Identification

Abstract

Dental records have been widely used as tools in forensic identification. With the vast volume of cases that need to be investigated by forensic odontologists, a move towards a computer-aided dental identification system is necessary. We propose a computer-aided framework for efficient matching of dental x-rays for human identification purposes. Given a dental x-ray with a marked region of interest (ROI), we search the database of x-rays (presumed to be taken from known individuals) to retrieve a closest match. In this work we use a slightly extended Weighted Sum of Squared Differences (SSD) cost function to express the degree of similarity/overlap between two dental radiographs. Unlike other iterative Least Squares methods that use local information for gradient-based optimization, our method finds the globally optimal translation. In 90% of the identification trials, our method ranked the correct match in the top 10% using a database of 571 images. Experiments indicate that matching dental records using the extended SSD cost function is a viable method for human dental identification.