Point cloud simplification method based on two-fold neighboring feature preservation

Three-dimensional point cloud models with high precision and density often have redundant data, resulting in an increase in computational time and hardware costs. To analyze and extract data more effectively, point cloud simplification is necessary. Existing algorithms are limited in the ability to preserve important features while considering the reconstruction error of the model. To address these issues, this paper proposes a point cloud simplification method based on two-fold neighboring feature preservation. The proposed method reduces the amount of data while retaining feature points, and extracting non-feature points based on the location of the neighboring feature points. The proposed algorithm, the curvature sampling method, the bounding box method and the random sampling method are compared. Experimental results show that the proposed method can minimize the surface and volume changes of the reconstruction model. Therefore, our method can be applied to models with different curvatures and can effectively avoid the hole phenomenon in the simplification process.