Abstract: To quickly detect the inconsistent areas between the denture machining model and the design model and guide the workers to quickly complete the denture correction,an Iterative Closest Point registration method based on the overlap rate of point clouds was proposed and a heat map was used to indicate the inconsistent areas of the denture machining model.Based on Iterative Closest Point(ICP) arithmetic,a bidirectional KD-tree was used to search the nearest points,a tolerance range was introduced to define the overlap points,and the overlap rate was used as the convergence condition for alignment instead of the mean square error.The results showed that the proposed method solved the problem of aligning the denture machining model with the design model and could quickly detect the discrepancy regions of the denture machining model for different types of prostheses.The overlap rate and registration efficiency were improved compared with the traditional ICP.

Key words: point cloud registration, iterative closest point, point clouds overlap, bidirectional KD-tree

摘要： 为了快速检测义齿加工模型与设计模型的不一致区域,以指导工人快速完成义齿修正,提出了一种基于点云重叠率的最近点迭代配准方法,并采用热力图的形式显示义齿加工模型不一致区域。在最近点迭代(ICP)算法基础上,使用双向KD-tree搜索近邻点,引入容差范围定义重合点,以重叠率代替均方误差作为收敛条件,进行配准。结果表明,所提方法解决了义齿加工模型与设计模型的配准问题,能够针对不同类型的义齿快速检测出义齿加工模型的差异区域,相比传统ICP,配准重叠率和效率均有提升。

关键词: 点云配准, 最近点迭代, 点云重叠, 双向KD-tree