Complex profile point cloud normal feature clustering hierarchical estimation method

Abstract

The invention provides a point cloud normal estimation method based on feature region clustering and grading in order to improve the accuracy of a normal estimation result of a complex profile sampling point cloud containing edges, sharp corners and other features, and belongs to the technical field of product reverse engineering. Performing clustering analysis on the point cloud according to thelocal flatness of the curved surface and the Bayesian information criterion, sequentially dividing the point cloud into flat, feature edge, edge sharp corner and other areas, identifying the feature type of the sample point, iterating the normal estimation result of the sample point in the flat area and propagating the normal estimation result to the adjacent feature area step by step, and obtaining a clustering result; wherein the normal estimation result of the characteristic sample point is consistent with the normal of the adjacent flat area sample point. The method can accurately estimatethe normal of the sample points in the characteristic area, effectively ensures the normal ambiguity of the sample points in the edge and sharp corner areas, and has an inhibition effect on data noise.

Description

technical field [0001] The invention provides a complex surface point cloud normal feature clustering and classification estimation method, which belongs to the technical field of product reverse engineering. Background technique [0002] The normal information of point cloud is an indispensable attribute of 3D model. The accuracy and consistency of normal estimation directly affect the accuracy of post-processing such as point cloud registration, streamlining and surface reconstruction. With the application of point cloud data in the fields of reverse engineering, industrial manufacturing, and medical visualization, the accuracy of normal estimation of complex surface sampling point clouds, especially point clouds with edge features, has been receiving much attention. [0003] At present, the most widely used normal estimation algorithm is the point cloud micro-section plane estimation method proposed by Hoppe et al. in the paper "Surface reconstruction from unorganized poi...

AI Technical Summary

Problems solved by technology

In the field of industrial design, the digital models of many products are composite surface models constructed based on modeling processes such as surface intersection, clipping, and chamfering. The reflected shape is regarded as a plane, which will lead to a large deviation between the estimated results of the normal and curvature of the sample points and the real situation.

Images