Point Cloud Simplification Algorithm for Oral Cavity Model based on Feature Significance Evaluation

Abstract

Point cloud simplification is an important pretreatment method for the surface reconstruction of oral cavity model with complex surface features and point cloud uneven sampling. The difficulty lies in how to accurately extract feature points and efficiently simplify redundant points to meet the requirements of high-precision reconstruction of oral cavity model. In this paper, we proposed a point cloud simplification algorithm for oral cavity model based on feature significance evaluation. First, point cloud down sampling is used to simplify the original points and k-neighboring points are searched by dynamic search cube method. Second, the feature significance evaluation formula is constructed by using feature measurement factors such as curvature, average projection distance and average spatial distance to extract the feature points. Finally, a hybrid segmentation method is used to segment remaining points, and the simplification rules and octree structure are used to realize the non-uniform simplification of the non-feature points. The experimental results show that the proposed algorithm can well retain the detail features of complex surfaces such as tooth contour and tooth structure, avoid the occurrence of holes in flat areas and the expansion of interdental holes. The algorithm has achieved good effects in terms of simplicity and accuracy.