A Framework of Point Cloud Simplification Based on Voxel Grid and Its Applications
Le Shi, Jun Luo
Abstract
As an information intensive 3-D representation, point clouds are usually characterized by extensive data, high redundancy, and uneven point density, which hinder their applications in many emerging fields. In order to solve the problems of large computation, feature disappearance, and reconstruction holes in the detection and 3-D reconstruction of complex surfaces, we propose a novel point cloud simplification framework based on the multi-feature fusion of voxel grid to achieve a balance between clear features and local uniformity in the down-sampling process. This effective internal control strategy improves the detection efficiency of the global region and avoids redundant computation. To verify the effectiveness of the proposed method, we simulated and validated it on the public datasets and compared it with others. The proposed down-sampling framework achieves excellent results in the applications of point cloud simplification, shape registration, and 3-D reconstruction. Finally, the framework is applied to the point cloud data simplification of the aero-engine turbine blade, and the advantages of the proposed method are verified by registration experiments.