Method for planning three-dimensional scanning viewpoint, device for planning three-dimensional scanning viewpoint, and computer readable storage medium

Patent metadata

Field	Value
Publication number	US-11776217-B2
Application number	US-201917431737-A
Country	US
Kind code	B2
Filing date	Jul 26, 2019
Priority date	Feb 20, 2019
Publication date	Oct 3, 2023
Grant date	Oct 3, 2023

Abstract

Disclosed in the embodiments of the present disclosure are a method for planning three-dimensional scanning viewpoint, a device for planning three-dimensional scanning viewpoint and a computer readable storage medium. After a low-precision digitalized model of an object to be scanned is acquired, viewpoint planning calculation is performed, on the basis of a viewpoint planning algorithm, on point cloud data in the low-precision digitalized model, and then the positions and line-of-sight directions of a plurality of viewpoints in space are calculated when a three-dimensional sensor needs to perform three-dimensional scanning on said object. Calculating viewpoints of a three-dimensional sensor by means of a viewpoint planning algorithm can effectively improve the accuracy and scientific nature of sensor posture determination, greatly improving the efficiency of viewpoint planning, and reducing the time consumed in the whole three-dimensional measurement process.

First claim

What is claimed is:

1. A method for planning three-dimensional scanning viewpoint, comprising:
acquiring a low-precision digitalized model of an object to be scanned, wherein the low-precision digitalized model is used to indicate spatial information of the object to be scanned;
performing a viewpoint planning calculation on point cloud data in the low-precision digitalized model according to a preset viewpoint planning algorithm; and
if it is determined that a preset algorithm termination condition is met, determining all calculated viewpoint information as a target viewpoint set of a three-dimensional sensor during overall scanning of the object to be scanned, wherein the viewpoint information comprises viewpoint position information and viewpoint direction information;
wherein the performing the viewpoint planning calculation on point cloud data in the low-precision digitalized model comprises:
estimating surface normals of the point cloud data in the low-precision digitalized model, to obtain point cloud normal vectors to a surface of the object to be scanned;
transforming the point cloud normal vectors to a spherical coordinate system, and grouping the point cloud normal vectors in the spherical coordinate system into a matrix of M lines×N columns;
collecting histogram statistics for a number of the point cloud normal vectors in each grouping region in the matrix; moving a window of a preset size cyclically on the matrix with each grouping region as a center; collecting statistics for a number of the point cloud normal vectors in the window in each movement, to determine a window with a largest number of the point cloud normal vectors; and then determining pointing directions ni=(xi, yi, zi) of the three-dimensional sensor based on a point cloud normal vector in the center of the window with the largest number of the point cloud normal vectors;
establishing a minimum bounding box for point clouds in the window with the largest number of the point cloud normal vectors, dividing the minimum bounding box based on a size of a field of view (FOV) space of the three-dimensional sensor, and determining three-dimensional coordinates of a center of the divided space as Ci=(xi, yi, zi), and
determining viewpoint information of the three-dimensional sensor in space according to ni, Ci, and a focusing distance D of the three-dimensional sensor.

2. The method for planning three-dimensional scanning viewpoint according to claim 1 ,
wherein before the performing the viewpoint planning calculation on point cloud data in the low-precision digitalized model according to the preset viewpoint planning algorithm, the method further comprises:
performing edge trimming on the low-precision digitalized model, and

wherein the performing the viewpoint planning calculation on point cloud data in the low-precision digitalized model according to the preset viewpoint planning algorithm comprises:
performing the viewpoint planning calculation on the point cloud data in the low-precision digitalized model, which has been subjected to edge trimming, according to the preset viewpoint planning algorithm.

3. The method for planning three-dimensional scanning viewpoint according to claim 2 , wherein a size of the window is determined based on a field of view (FOV) angle of the three-dimensional sensor and a visibility constraint, and the visibility constraint is that a product of the point cloud normal vector and a viewpoint direction of the three-dimensional sensor is less than zero.

4. The method for planning three-dimensional scanning viewpoint according to claim 2 , wherein the establishing the minimum bounding box for the point clouds in the window with the largest number of the point cloud normal vectors comprises:
acquiring feature vectors of the point clouds in the window with the largest number of the point cloud normal vectors by means of Principal Component Analysis (PCA), and re-defining coordinate axes based on the acquired feature vectors; and
determining a centroid of the point clouds in the window with the largest number of the point cloud normal vectors in the re-defined coordinate axes, and establishing the minimum bounding box for the point clouds in the window with the largest number of the point cloud normal vectors according to the re-defined coordinate axes and the centroid.

5. The method for planning three-dimensional scanning viewpoint according to claim 1 , wherein the determining that the preset algorithm termination condition is met comprises:
determining, based on a number Npresent of point clouds to be calculated currently, that the preset algorithm termination condition is met, wherein Npresent denotes a number of point clouds to be calculated that are determined from point clouds remaining after the previously calculated point clouds are deleted.

6. The method for planning three-dimensional scanning viewpoint according to claim 5 , wherein the determining, based on the number Npresent of point clouds to be calculated currently, that the preset algorithm termination condition is met comprises:
determining a point cloud ratio p=Npresent/Ntotal according to the number Npresent of the point clouds to be calculated currently and an initial number Ntotal of point clouds; and
if p is less than a preset ratio threshold, determining that the preset algorithm termination condition is met.

7. The method for planning three-dimensional scanning viewpoint according to claim 6 , wherein a size of the window is determined based on a field of view (FOV) angle of the three-dimensional sensor and a visibility constraint, and the visibility constraint is that a product of the point cloud normal vector and a viewpoint direction of the three-dimensional sensor is less than zero.

8. The method for planning three-dimensional scanning viewpoint according to claim 6 , wherein the establishing the minimum bounding box for the point clouds in the window with the largest number of the point cloud normal vectors comprises:
acquiring feature vectors of the point clouds in the window with the largest number of the point cloud normal vectors by means of Principal Component Analysis (PCA), and re-defining coordinate axes based on the acquired feature vectors; and
determining a centroid of the point clouds in the window with the largest number of the point cloud normal vectors in the re-defined coordinate axes, and establishing the minimum bounding box for the point clouds in the window with the largest number of the point cloud normal vectors according to the re-defined coordinate axes and the centroid.

9. The method for planning three-dimensional scanning viewpoint according to claim 5 , wherein a size of the window is determined based on a field of view (FOV) angle of the three-dimensional sensor and a visibility constraint, and the visibility constraint is that a product of the point cloud normal vector and a viewpoint direction of the three-dimensional sensor is less than zero.

10. The method for planning three-dimensional scanning viewpoint according to claim 5 , wherein the establishing the minimum bounding box for the point clouds in the window with the largest number of the point cloud normal vectors comprises:
acquiring feature vectors of the point clouds in the window with the largest number of the point cloud normal vectors by means of Principal Component Analysis (PCA), and re-defining coordinate axes based on the acquired feature vectors; and
determining a centroid of the point clouds in the window with the largest number of the point cloud normal vectors in the re-defined coordinate axes, and establishing the minimum bounding box for the point clouds in the window with the largest number of the point cloud normal vecto

Assignees

• Univ Shenzhen

Inventors

• Liu Xiaoli
• Liao Yifan
• Chen Hailong
• Peng Xiang
• Tang Qijian

Classifications

• G06T19/00 Primary

  Manipulating three-dimensional [3D] models or images for computer graphics · CPC title

• G06T7/70

  Determining position or orientation of objects or cameras (camera calibration G06T7/80) · CPC title

• G06T7/85

  Stereo camera calibration · CPC title

• G06T17/00

  Three-dimensional [3D] modelling for computer graphics · CPC title

• H04N13/221

  using the relative movement between cameras and objects · CPC title