Fast three-dimensional point cloud searching and matching method

Abstract

The invention relates to a fast three-dimensional point cloud searching and matching method, which comprises the steps of main shaft registration, measuring shaft generation, low-precision searching, high-precision searching and primary iteration result generation. The main shaft registration link is responsible for the defining point cloud direction and completes the initial registration, the measuring shaft generation link provides the searching range in the current iteration for moving point clouds, the low-precision searching adopts an ergodic method for looking for the optimal registration posture of the moving point clouds on the current measuring shaft, the high-precision searching is responsible for obtaining better optimal registration posture at smaller step length searching position errors on the basis of the low-precision searching, and the primary iteration result generation link is responsible for generating the final results of the primary iteration according to the measuring results in each position. The invention starts from the geometrical shape of the point clouds, and completes the ergodic searching at different precisions through the mutual movement and rotation between two point clouds to find the optimal matching position, the algorithm has good fault-tolerance performance, and the invention is suitable for any three-dimensional digital instrument to obtain the searching matching of the three-dimensional point clouds.

Description

technical field [0001] The invention relates to a fast three-dimensional point cloud search and matching method, which belongs to the technology of remote sensing and surveying and mapping, computer vision and pattern recognition, and is applicable to three-dimensional point cloud data obtained by any three-dimensional digitizer. Background technique [0002] 3D point cloud registration has always been an issue of great concern in the fields of computer vision and pattern recognition, remote sensing and surveying and mapping. Most of the existing registration methods are based on the ICP algorithm to iteratively find the closest point between two point clouds, so as to achieve point cloud registration. If there are similar structures in the scene, iterations tend to fall into local minima, and mismatches occur, which leads to the failure of the entire registration. Contents of the invention [0003] The technology of the present invention solves the problem: overcomes the...

AI Technical Summary

Problems solved by technology

If there are similar structures in the scene, iterations tend to fall into local minima, and mismatches occur, which leads to the failure of the entire registration

Images