A global optimization and global registration method for multi-view 3D laser point cloud

Abstract

The present invention provides an overall registration method for automatic global optimization of known multi-view laser point cloud registration initial values, which is characterized in that firstly, a global optimization global adjustment model for multi-view laser point cloud registration is established and deduced in detail, and by estimating The density of the point cloud automatically detects all two-view point clouds with a certain degree of overlap; then uses the K-D tree to search for similarly named points in the two-view point cloud with a certain degree of overlap, and substitutes them into the global optimization adjustment model as observations In this method, through the iterative adjustment calculation, the optimal rotation and translation transformation parameters of the multi-view laser point cloud are obtained at the same time, so as to complete the overall accurate registration of the multi-view 3D laser point cloud. The present invention focuses on improving the overall registration accuracy of multi-view 3D laser point clouds, and can simultaneously process disordered and scattered multi-view 3D laser scanning point clouds, and conduct registration experiments through actual multi-view 3D laser point cloud data. On the basis of registration efficiency, the overall registration accuracy is effectively improved.

Description

technical field [0001] The invention belongs to the technical field of photogrammetry and computer vision, and in particular relates to a multi-view three-dimensional laser point cloud global optimization overall registration method. Background technique [0002] In recent years, 3D laser scanning measurement technology has developed rapidly and has been widely used in reverse engineering, industrial measurement, and digital protection of cultural relics. In practical applications, in order to obtain a complete 3D point cloud on the surface of the object, it is necessary to scan the measured object from multiple angles, and then it is necessary to register the multi-view laser point cloud into a unified coordinate system, so the registration accuracy will directly affect the final object. 3D reconstruction accuracy. Therefore, it is of great practical significance to globally optimize the overall registration of the 3D laser point cloud after the initial registration. [0...

AI Technical Summary

Problems solved by technology

This method has a high degree of automation, but it needs to obtain 3D point cloud data on the surface of the object with the help of a rotating platform, and at the same time needs to calibrate the relative positional relationship between the rotating platform and the scanner, so the applicability is relatively low

Liu Jun proposed a method for 3D registration of multi-viewpoint cloud based on sequence iteration. This method effectively suppresses the cumulative error of sequence registration, but the registration process is more manual and inefficient.

Images