An approach for single tooth classification and identification

This paper presents the study of the distinctiveness of single human tooth and the development of three dimensional (3D) single-tooth identification scheme which is useful when only very partial jaws with teeth are available. Two types of single-tooth samples are studied: tooth crown segmented from scanned dental casts and entire single tooth (crown and root) reconstructed from cone beam CT images. Eigenteeth feature and K nearest neighbourhood (KNN) classifier are applied to classify posterior (molar, premolar) and anterior (canine and incisor) tooth types. The iterative closest point algorithm (ICP) is then applied to match post-mortem (PM) and ante-mortem (AM) tooth data, with the former being the sample to be matched, and the latter representing captured complete tooth geometry kept in a database.