A method for quickly generating building cross-sections based on point cloud data

Abstract

The invention provides a method for quickly generating a cross-section diagram of a building based on point cloud data, which comprises the steps of measuring point cloud data of the building, calculating coordinate transformation parameters, and performing point cloud registration and denoising; cutting a cross section of the building by a clipping cube, obtaining edge point cloud data, and forming two-dimensional plane data P(xi, yi)<T>; performing translation and amplification on the point cloud data, and extracting collinear point cloud data Pk1(xi, yi)<T>; removing interference points inthe collinear point cloud data Pk1(xi, yi)<T> by using an angle [Theta]k and polar radius [rho]k to obtain collinear point cloud data Pk2(xi, yi)<T>; sorting and segmenting collinear point cloud, finding out a segment with the maximum number of points to reserve, obtaining point cloud line segment data Pk3(xi, yi)<T>, and deleting the point cloud line segment data Pk3(xi, yi)<T> from the two-dimensional plane data P(xi, yi)<T>; executing the above steps in a circulating manner to extract point cloud segments, and terminating the circulation if the point number of the point cloud line segmentsis less than a point number threshold t4; thinning the point cloud line segment data Pk3(xi, yi)<T> to obtain Pk4(xi, yi)<T>, then performing accurate straight line fitting by adopting a selecting weight iteration based least square method to obtain endpoint coordinates of each line segment; and connecting the endpoint coordinates of the line segments end to end to obtain a cross-section diagram of the building.

Description

technical field [0001] The invention belongs to the technical field of three-dimensional laser scanning, and in particular relates to a method for quickly generating building cross-sections based on point cloud data. Background technique [0002] In engineering applications, it is often necessary to obtain a cross-sectional view of a building. The traditional cross-sectional view of a building is to use a total station and other equipment to collect the three-dimensional coordinates of each independent point, and then draw it by software or manually. This method has a large field workload, low efficiency and The internal processing process is cumbersome and only partial information can be obtained. [0003] 3D laser scanning is a new technology developed in recent years. It has become an effective means of spatial 3D data acquisition with the advantages of high precision, massive, fast, and non-contact measurement. The acquired point cloud data is a series of discrete point...

AI Technical Summary

Problems solved by technology

[0002] In engineering applications, it is often necessary to obtain a cross-sectional view of a building. The traditional cross-sectional view of a building is to use a total station and other equipment to collect the three-dimensional coordinates of each independent point, and then draw it by software or manually. This method has a large field workload, low efficiency and The internal processing process is cumbersome and only partial information can be obtained

[0003] 3D laser scanning is a new technology developed in recent years. It has become an effective means of spatial 3D data acquisition with the advantages of high precision, massive, fast, and non-contact measurement. The acquired point cloud data is a series of discrete point, in order to use point cloud data to obtain a building cross-section, it is first necessary to detect each line segment in the cross-section, the traditional line detection method is mainly the Hough transform, and the direct use of the Hough transform for point cloud line detection can only achieve Limited accuracy, especially when affected by noise, if the accuracy requirement is too high, the detection may fail

The least squares method can obtain an accurate straight line in the sense of mean square error, but it is easily disturbed by noise points. If the least squares fitting is performed directly, it will be interfered by strong noise, and the purpose of obtaining high-precision straight line parameters cannot be achieved.

Images