High-precision point cloud normal calculation method

Abstract

The invention belongs to the technical field of point cloud processing, and particularly discloses a high-precision point cloud normal calculation method, which comprises the following steps that point cloud data is read; a three-dimensional point cloud containing N points is set as P, the ith point in the three-dimensional point cloud P is marked as Pi (xi, yi, zi), three scales k1, k2 and k3 of the Pi are determined as neighbor points in a spatial neighborhood, the point cloud is traversed, and an initial normal vector corresponding to each scale of all the points in the three-dimensional point cloud P is determined; neighborhood point normal information is used for replacing the invalid normal, a minimum spanning tree algorithm is adopted to enable the normal directions of all neighborhood points to be consistent, and the final normal directions of all points in the P are determined. According to the method, a mechanism that multi-scale information is adopted to calculate and fuse the normal direction is proposed for the first time, the unstable influence caused by parameter K selection is effectively avoided, the influence of noise is inhibited, details are better kept at the boundary part, and the normal calculation problem of noisy point cloud data is effectively solved.

Description

technical field [0001] The invention belongs to the technical field of point cloud processing, and in particular relates to a high-precision point cloud normal calculation method. Background technique [0002] With the advancement of 3D sensor technology, the acquisition of 3D point cloud becomes easier and easier. The information described by simply using the three-dimensional space coordinates of points is very limited. The normal direction is the most widely used and most important feature of point clouds, and is widely used in data analysis, registration, segmentation, etc. [0003] In related technologies, the 3D point cloud data of the collected image usually has noise erosion. In terms of hardware, due to the accuracy error of the depth camera or 3D laser scanner, noise will appear; in terms of software, the algorithm for generating point cloud In the process, due to the error of angle measurement or the inaccuracy of floating point number operation, the point may be...

AI Technical Summary

Problems solved by technology

The selection of the nearest neighbor parameter K in this method is very critical. If it is too large, it will cause the loss of details, if it is too small, it will be sensitive to noise, and the point cloud data itself often contains a lot of noise. Therefore, this method is used to estimate the points in the point cloud. It is easy to introduce a large error when the normal vector is normal vector, resulting in a large amount of noise in the point cloud data

Images