Dental biometrics: Human identification based on dental work information

Abstract

Dental biometrics is used in forensic dentistry to identify or verify persons based on their dental radiographs. This paper presents a method for human identification based on dental work information. The proposed method works with three main processing steps: segmentation (feature extraction), creation of a dental code, and matching. In the segmentation step, seed points of the dental works are detected by thresholding. The final segmentation is obtained with a snake (active contour) algorithm. The dental code is defined from the position (upper or lower), the size of the dental works, and distance between neighboring dental works. The matching stage is performed with the Edit distance (Levenshtein distance). The costs for the insertion, deletion and substitution operations were adapted to make the matching algorithm more sensitive. The method was tested on a database including 68 dental radiographs and the results are encouraging.