Point Cloud Simplification Algorithm Based on Hausdorff Distance and Local Entropy of Average Projection Distance

Abstract

The high precision digital dental model obtained by intraoral optical scanner poses great challenges to processing speed and storage space of computers when performing tasks such as data storage, transmission and management. A simplification algorithm based on Hausdorff distance and local entropy of average projection distance is proposed to meet the practical application requirements of digital impression technology. After the edge points are extracted and the interdental points are sampled, the proposed algorithm used octree structure to segment the remaining points. According to the average Hausdorff distance of all points in each subcube and the average Hausdorff distance of the remaining points, the corresponding obtained evaluation ratio is compared with different thresholds of the multi-level simplification criterion, so as to determine the feature degree of the subcube. The point cloud simplification factor and the local entropy of average projection distance are used to preserve the corresponding number of feature points in the feature subcube. According to the local point cloud distribution, all points in the non-feature subcube can be replaced by a certain number of points closest to the center of all points. The experimental results show that the proposed algorithm can well preserve many detail features of the dental model and effectively avoid the appearance of holes in non-feature regions. The proposed algorithm is superior to other comparison algorithm in terms of simplification effect and simplification error.