A novel 3D dental identification framework is presented. The objective is to develop a methodology to enable computer-automated matching of complex dental surfaces with possible missing regions for human identification. Thus far, there is no reported attempt at 3D dental identification given partially available dental casts or impressions. This approach overcomes a number of key hurdles in traditional 2D methods. Given the 3D digital form of a dental cast surface, the developed method will facilitate the search for the closest match in the database of digitized dental casts. A salient curvature matching algorithm (SCM) is proposed for pose estimation which includes algorithms for feature extraction, feature description and correspondence. The feature point extraction algorithm could extract more salient features and the correspondence algorithm is more robust for pose estimation compared to known works. Experimental results show 85.7% hit rate at rank-1 accuracy based on matching of 7 partial sets to a database of 100 full sets in significantly reduced retrieval time. The hit rate increases to 100% with parameter adjustment. This work aims to enable computer-aided 3D dental identification and the proposed method could be adjunctively used with the traditional 2D dental identification method, as the available dental source for identification is still primarily 2D radiographs. Limitations of the methodology and future directions in matching highly fragmented and partial dental surfaces are discussed.

This content is only available via PDF.
You do not currently have access to this content.