The 3D scanning device can capture millions of points with the excellent geometrical precision except for large amount of redundant ones, which could bring some difficulties for the subsequent digital geometrical processing (DGP), so the simplification of the point cloud has become a considerable study in point cloud applications. Given this problem, we propose a novel approach, which could decrease the geometrical error by partitioning the surface into many patches with some similar geometrical features before using the adaptive hierarchical space partition (AHSP) approach, in order to improve the AHSP simplification. We also experiment on three models and do comparative analysis. Fortunately, the results prove that our algorithm can make the anisotropy feature in the surface of the models described explicitly, the geometrical error decreased by 15.8 percent, and the simplification rate kept up with other approaches. In addition, it can provide the high quality models for the 3D digital model editing, such as the geometrical modeling, the point cloud blending.
Evolving Non-Dominated Parameter Sets for Computational Models from Multiple Experiments