Point cloud feature point detection method and cloud point feature extraction method

Abstract

The invention provides a point cloud feature point detection method and a cloud point feature extraction method. The checking method comprises the following steps that: taking one data point in pointcloud data as a sphere center to establish a local spherical coordinate frame, and finding k pieces of neighbourhoods through a k nearest neighbor algorithm; independently connecting the sphere centerwith each nearest neighbor point to obtain k pieces of line segment groups; according to the horizontal projection included angles of the line segment groups, sorting k pieces of line segment groups,checking each line segment group through a Laplace operator, and determining whether the data point is a feature point or not; and according to the above steps, processing each data point in the cloud point data, and obtaining all feature points of the cloud point data. The point cloud feature extraction method comprises the following steps that: according to the above detection method, determining feature points, and recording the regional connection information of each feature point; and determining a sequential connection relationship between the feature points, connecting the feature points, and forming a segmented feature polygon to realize region segmentation. The amount of labeled feature points is small, feature points are extracted orderly, and feature lines are conveniently reconstructed.

Description

technical field [0001] The invention relates to point cloud model storage, in particular, to a point cloud feature point detection method and a point cloud feature extraction method. Background technique [0002] 3D laser scanning technology is widely used in mechanical design, cultural relic protection, medicine, architectural measurement and other fields. More and more physical samples and CAD models are stored in point cloud models. Point cloud models often only have spatial coordinate information, without geometric attributes such as surface normal and curvature, and topological structures such as triangular meshes and model parameters. There are various geometric structures such as feature points, characteristic curves, and surfaces in the model. Separating various geometric structures cannot reasonably segment the surface slices of the point cloud, and it is difficult to directly reconstruct the physical sample model from the point cloud data as a whole. Even so, it al...

AI Technical Summary

Problems solved by technology

Feature point detection involves point geometric attribute estimation. Since there is no topological structure and other factors, point cloud geometric attribute estimation is by finding the k-nearest neighbor point of a given point, fitting the nearest neighbor point as a paraboloid, and approximating the point curvature with the paraboloid curvature, or Use methods such as PCA to estimate the normal direction of the point by generating a covariance matrix, or estimate the dihedral angle by generating a triangular mesh or Voronoi mesh of the point cloud to help feature point detection. The geometric quantities estimated by these methods are not high in accuracy. Sharp feature errors are especially noticeable

In addition, the comparison between point geometric quantities is carried out between any two points in the neighborhood. As long as the deviation is greater than a given threshold, the point is considered as a feature point. The number of marked feature points is large, which brings difficulties to feature line reconstruction.

[0004] Feature line reconstruction is to use the extracted feature points, regional clustering of non-feature points and connection information between feature points and regions to connect feature points into curve segments. The order of feature points is very critical in the reconstruction stage. Before the feature line is unknown, any Feature points may be located on multiple feature lines. Existing methods connect feature lines by comparing distances or generating minimum spanning tree paths. Errors may occur in feature line connections

Images