Point cloud single-point alignment method based on ridge-valley feature and depth feature descriptors

Abstract

The invention discloses a point cloud single-point alignment method based on ridge-valley feature and depth feature descriptors, and the method comprises the steps: S1, carrying out the feature extraction of a point cloud, and obtaining ridge-valley feature points; s2, constructing a local coordinate system; s3, dividing grids on the local coordinate system to generate regular grid data; S4, constructing a lightweight network PFNet; s5, generating feature descriptors in the regular grid data PFNet, and searching the feature descriptors by using a KD tree to obtain matching point pairs; s6, generating a candidate solution by utilizing the local coordinate system mapping of the matching point pair; and S7, filtering the candidate solutions obtained in the step S6 by using an RANSAC algorithm. According to the method, data alignment depending on a single feature point is achieved, a good alignment effect can be achieved under the point cloud lacking complex features, and high robustness is achieved for noise, outliers, non-uniform sampling and the like.

Description

technical field [0001] The invention relates to the technical field of point cloud alignment, in particular to a point cloud single-point alignment method based on ridge-valley features and depth feature descriptors. Background technique [0002] Point cloud is a data representation of the surface of a three-dimensional object, generally obtained by a laser scanner, and it is widely used in reverse engineering, biomedical, virtual technology and other fields. Point cloud alignment is a necessary step to process point clouds and realize the digitization of physical objects. It can be divided into two processes: coarse alignment and fine alignment. Among them, the coarse alignment is used to roughly put together two point clouds that are far apart; the role of the fine alignment is to further optimize the point cloud pose on the basis of the coarse alignment and improve the accuracy of the alignment. Although related issues have been extensively studied, due to the complex ch...

AI Technical Summary

Problems solved by technology

However, there are still some problems in the above method, such as the use of 3D data for training takes a lot of time; the randomly selected key points have poor repeatability, etc.

Images