Point cloud splicing method and system

Abstract

The invention provides a point cloud splicing method and system. The method comprises the following steps: inputting two groups of point clouds and a preset splicing precision; carrying out initial splicing on the input two groups of point clouds by utilizing match feature points to obtain two groups of initial feature point sets; removing abnormal points in the two groups of initial feature point sets according to the preset constraint conditions to obtain finally-determined two groups of feature point sets; calculating an objective function corresponding to the two groups of feature point sets determined above, and carrying out least square iteration on the two groups of feature point sets determined above to obtain a group of transformed point clouds; and when the precision of the objective function is smaller than the preset splicing precision, carrying out precise splicing on the transformed point clouds and the other group of point clouds in the feature point sets. The point cloud splicing method and system can realize quick and precise point cloud splicing.

Description

technical field [0001] The present invention relates to a point cloud processing method and system, in particular to a point cloud splicing method and system. Background technique [0002] In the point cloud scanning technology, to scan a complete product, the product needs to be scanned at different locations, and then the point clouds scanned at different locations are spliced ​​to obtain a complete point cloud of the product. When the existing point cloud stitching technology performs point cloud stitching, it generally performs initial registration based on landmark points, image feature points, and boundary feature points, and then uses the Iterative Closest Point (ICP) algorithm to perform point cloud accuracy. stitching. However, the ICP algorithm is easily affected by measurement noise. When the point cloud to be stitched contains a lot of noise, even if the initial value obtained by the initial registration is good, the ICP algorithm cannot obtain an accurate stitc...

AI Technical Summary

Problems solved by technology

However, the ICP algorithm is susceptible to measurement noise. When the point cloud to be stitched contains a lot of noise, even if the initial value obtained by the initial registration is good, the ICP algorithm cannot obtain an accurate stitching result.

In addition, the calculation time of the ICP algorithm is long, the cumulative error is large, and the density of the point cloud at the stitching point is high, and the stitching traces after stitching are obvious

Images