Method of identifying sharp geometric edge points based on normal consistency

Abstract

The invention discloses a method of identifying sharp geometric edge points in 3D point cloud based on normal consistency. The method is specifically implemented according to the following steps: calculating the normal direction and normal consistency of each point in 3D point cloud; looking for local extreme points with normal consistency in the neighborhood range of each point; judging whether there is a sharp edge nearby according to the number of local extreme points in the neighborhood range, and if there is a sharp edge, further classifying all the points in the neighborhood range according to the normal distribution, and finally judging whether each point is a sharp edge point according to whether the neighborhood center point is on the classification edge; and using the method above to locate all sharp edge points in the point cloud. The edge point detection method of the invention can adapt to any change of edge shape. No model assumption is needed for the intersecting surfaces forming a sharp edge. Straight and curved edges as well as corners formed by intersecting surfaces can be detected. Moreover, all kinds of sharp geometric edges can be identified from 3D point cloud.

Description

technical field [0001] The invention relates to a sharp geometric edge point recognition method utilizing normal direction consistency, and belongs to the field of three-dimensional point cloud data processing. Background technique [0002] Three-dimensional laser scanning is a three-dimensional space imaging technology that has developed rapidly in recent years. It uses a two-axis servo motor to drive a laser rangefinder to continuously sample the target surface of the scanning scene at equal intervals, and output a three-dimensional image that is completely consistent with the geometry of the three-dimensional scene. point cloud. This technology is being more and more widely used in surveying and mapping, archaeology, machining and manufacturing, three-dimensional printing, robotics, unmanned driving and other fields. [0003] Geometric edge points refer to those points distributed near sharp geometric edges in the 3D point cloud. Geometric edges can not only clearly out...

AI Technical Summary

Problems solved by technology

[0005] The existing edge point recognition methods (belonging to the second category) based on normal mutations and curvature maximums are all sensitive to noise, and a large noise point may cause adjacent points to be misidentified as edge points; The sensitivity to noise also makes it difficult for these methods to identify less obvious geometric edges, such as geometric edges formed by two intersecting surfaces with an angle close to 180°

Images