Tooth Segmentation of Dental Mesh Based on Improved Region Growing

Abstract

Computer-aided design (CAD) systems have gained more and more concern in orthodontics over the last decades. Aimed at tooth segmentation task of three dimension (3D) dental models in orthodontic CAD systems, we propose an efficient and robust framework based on geometric features. To make region growing algorithm flexible and apply it to 3D tooth segmentation, we extend the region similarity description from single-parameter to multi-parameter. A parameter adaptive method is designed to get rid of manual parameter adjustment and raise efficiency. Moreover, we provide a multi-level label optimization algorithm for segmentation refinement. Some experiments are conducted with various dental models from real cases to validate various algorithmic components. The results demonstrate that our pipeline is superior on efficiency and robustness.