A point cloud stitching method for flat parts based on multi-dimensional space invariant features

Abstract

The invention discloses a point cloud splicing method of flat-plate parts based on multi-dimensional space invariant features. Use the nearest neighbor iterative algorithm to calculate the point cloud rotation matrix and multiple sets of sampling radii. For different sampling radii, use the point cloud texture and point cloud density features to calculate the covariance matrix of each point at different scales and construct a covariance description. character; define the multi-scale flow distance between the two covariance descriptors; determine the matching point pairs in the two point clouds according to the flow distance and roughly register to obtain the translation matrix; then perform the nearest neighbor point iterative algorithm to obtain the rotation matrix; obtain Finally transform the matrix to complete the splicing. The present invention defines a covariance descriptor for each point in the point cloud, utilizes the density feature and texture feature with spatial invariance characteristics at a certain point, and realizes the automatic splicing of the point cloud of flat-plate parts, which is suitable for plate-shaped parts with hole features. Stitching of part point clouds.

Description

technical field [0001] The invention relates to the field of post-processing of three-dimensional point cloud data, and mainly relates to a method for splicing point clouds of flat-plate parts based on multi-dimensional space invariant features. Background technique [0002] In industrial production, flat-plate parts as cover parts, base plates or base parts are widely used. In order to ensure the production quality, it is necessary to test the forming quality of such flat parts. In recent years, as a non-contact point cloud acquisition method with high measurement accuracy, structured light 3D measurement technology has developed rapidly. More and more companies have begun to use this technology to reconstruct the structural dimensions of formed parts in 3D to Check whether the forming of parts meets the design requirements. Limited by the size of large flat parts, structured light 3D measurement equipment can often only measure from a single perspective. Since the point...

AI Technical Summary

Problems solved by technology

It can be seen that the artificial assisted method mainly has the following problems in the mosaic of multiple point clouds: First, the assisted mosaic method of artificially marked points has high requirements for the setting of the shape, size, and position of the marked points, and destroys the Integrity of the surface of the object; second, the positioning device is used to determine the position transformation matrix between multi-view point clouds. The stitching result is affected by the accuracy of the positioning device, and the equipment is large in size and poor in operation flexibility.

However, the above splicing method has the problem that the point cloud surface information used is less or the information is not easy to obtain, and the stability is poor, so it is difficult to splice large flat parts with few features

Images