A Research On 3D Tooth Segmentation Technique Based On Depth Image And Projection Image

With the development of computer technology,the precision medicine in our country is developing rapidly.Computer-aided orthodontic systems can achieve computer-aided3 D diagnosis,customized personalized treatment plans,and visualization of the teeth straightening process.Its intuitive,precise,and personalized advantages make the orthodontic process more efficient and faster.Teeth segmentation is the most important step in computer-aided orthodontic systems.Based on the practical demand of orthodontics,this thesis presented an automatic segmentation algorithm for teeth and dentition.The algorithm can efficiently and accurately segment individual teeth,improving the efficiency of oral treatment.This thesis adopts a region growing-based method for segmenting dental arches and teeth.Firstly,the initial boundary region is identified by edge detection,and then the seed point set and stop condition of region growing algorithm are selected on this basis,and region growth is used for segmentation.The work in this thesis is mainly divided into the following three parts:(1)initial boundary detection based on depth images and projection images.The formula for coordinate conversion of data points between 3D and 2D is determined.Depth images are generated using the distance from data points to the dental arch surface,while projection images are generated using curvature and distance density of data points.Edge detection based on wavelet transforms is performed on the depth and projection images,and their results are combined to obtain the initial boundary.(2)Region growing algorithm based on multi-objective optimization.For seed point selection,this thesis uses the initial boundary for coarse-grained dental arch recognition and selects seed points that meet certain conditions on that dental arch.For stop conditions,this thesis analyzes the characteristics of the initial boundary point set,uses multiobjective optimization to select curvature and density thresholds as local stop conditions,and uses the initial boundary in 2D as the overall stop condition.(3)Implementation of tooth segmentation and its application.In this thesis,dentition segmentation and tooth segmentation are realized according to the proposed region growth algorithm,and an application system based on tooth segmentation is realized.Users can automatically segment dentition or teeth through this system,analyze arch morphology,occlusal plane analysis,WALA ridge analysis and other data,and save the analysis data.In this thesis,the accuracy of the proposed tooth segmentation algorithm can reach more than 95% through systematic experiments,and it has good applicability in data labeling.