中文摘要：

针对三维牙颌模型由于点云重建导致齿间区域融合的问题,提出一种能避开融合区域的牙齿分割算法。根据曲率筛选和邻接信息过滤得到特征区域,采用改进的形态学算法获得边界线,利用几何信息让边界线避开融合区域,通过改进的区域生长法把牙齿精确地分离出来。实验结果表明,使用该方法能够对不同牙齿模型进行正确分割,速度较为迅速,因避开了大部分融合区域,得到的单颗牙齿模型有助于牙齿矫形后续操作。

英文摘要：

For the existence of region fusion between teeth caused by point cloud reconstruction,an improved tooth segmentation approach was applied to avoid the fusion region.Feature region was selected using mean curvature and connectivity information.Boundaries were got by using improved morphologic algorithm.Boundaries were put to avoid fusion area by using geometry information.The teeth were segmented using an improved region growing method.The experimental results show that using the proposed algorithm can get correct segmentation results and it has quick speed,and the single teeth model obtained avoids most of the fusion regions,which helps conducting subsequent operations of dental orthopedics.