Study On The Segmentation And Deformation Of 3D Dental Model

The virtual orthodontic system is the software that outputs the braces model.The system lesigns the entire treatment plan through the reconstruction,tissue segmentation and eformation of the three-dimensional dental model of the patients,and customizes braces for he patient in different periods.However,there are still many shortcomings in this system,such as low precision of tooth segmentation,serrated teeth,slow segmentation speed and more manual intervention.Gingival deformation is not high simulation,deformation speed is slow and other problems.This topic focuses on the two modules of tooth and jaw model segmentation and gingival soft tissue deformation simulation in the virtual orthodontic system.The main research contents and innovation points are as follows:(1)In order to simplify the operation of seeds selection in tooth segmentation of 3D dental model,an automatic seeds picking method based on 2D projection of occlusial plane was proposed.Firstly,2D projection occlusial plane images of 3D dental model were captured,and then the 2D seeds data set for tooth segmentation was established by labeling according to the shape and category of teeth.Then,we built a prediction network of teeth seeds based on YOLO-v4 to realize the prediction of teeth position as well as the recognition of teeth categories.Finally,according to the statistical optimal seeds position,the two-dimensional seeds are calculated and mapped back to the three-dimensional space by the reverse projection transformation method to realize the final picking up of the three-dimensional seeds.Furthermore,combined with the results of dividing line detection in this project,the automatic segmentation of 3D dental model was realized.The experimental results show that the mean average precision of teeth categories recognition on the self-established data set is 93.882%,and the average detection time was 0.1884s.The proposed method has high accuracy and real-time performance,which significantly reduces the burden of human-computer interaction in dental model segmentation.(2)It is an important pretreatment of virtual orthodontic system to accurately segment the teeth from the model of teeth and jaws.In the present methods,most of the triangular-face segments are calculated and processed.The cutting plane and the cutting line are automatically constructed according to the position of seed points.Locking the detection range by looking for the position with the greatest negative curvature on the cutting line.Finally,the method is used to detect the dividing line according to the curvature and Angle information.Combined with the automatic seeds picking method of this project,the automatic segmentation of dental model is realized.The experimental results show that this method has strong adaptability to various malformed tooth model,and can achieve accurate tooth segmentation and greatly improve the segmentation efficiency.(3)In order to solve the problem that the boundary of gingiva is wrinkled after deformation in orthodontic simulation,we designed a simulation method in a wrapping manner for virtual gingiva based on the traditional Mass-spring model.Firstly,the transition boundary point between teeth and gums was selected as the active point in the deformation process,and the position of the active point was determined by wrapping it on the tooth surface after the correction operation.Then,in order to solve the problem of missing attachment point when the position of the active point is determined,we added a segment of tooth root based on the method of tooth contour curve fitting.Finally,we used the sum of surface curvature values to compare the experimental results.The experimental results show that the proposed method can effectively alleviate the problem of boundary wrinkles after gingival deformation in the case of large correction range,and improve the authenticity of the virtual orthodontic system.