Construction of three-dimensional road network map

Patent metadata

Field	Value
Publication number	US-12423912-B2
Application number	US-202318180670-A
Country	US
Kind code	B2
Filing date	Mar 8, 2023
Priority date	Mar 15, 2022
Publication date	Sep 23, 2025
Grant date	Sep 23, 2025

Abstract

A method for constructing a three-dimensional road network map is disclosed. The method includes: extracting a two-dimensional road based on a satellite remote sensing image; skeletonizing the two-dimensional road to obtain a two-dimensional road network map; and determining a height of each point in the two-dimensional road network map based on an elevation of three-dimensional point cloud data, to obtain the three-dimensional road network map.

First claim

What is claimed is:

1. A method for constructing a three-dimensional road network map, the method comprising:
extracting a two-dimensional road based on a satellite remote sensing image;
skeletonizing the two-dimensional road to obtain a two-dimensional road network map; and
determining a height of each point in the two-dimensional road network map based on an elevation of three-dimensional point cloud data, to obtain the three-dimensional road network map, wherein the determining the height of each point in the two-dimensional road network map based on the elevation of the three-dimensional point cloud data, to obtain the three-dimensional road network map comprises:
obtaining a plane point cloud based on the three-dimensional point cloud data;
determining, for a point in the two-dimensional road network map, one or more associated points of the point in the plane point cloud; and
determining a height of the point according to elevation of the one or more associated points, to obtain the three-dimensional road network map.

2. The method according to claim 1 , wherein the skeletonizing the two-dimensional road to obtain the two-dimensional road network map comprises:
skeletonizing the two-dimensional road, and determining a skeleton node segment based on skeleton nodes in the skeletonized two-dimensional road; and
converting, for the skeleton node segment, a pixel coordinate into a geographic coordinate, to obtain the two-dimensional road network map.

3. The method according to claim 2 , wherein the converting, for the skeleton node segment, the pixel coordinate into the geographic coordinate, to obtain the two-dimensional road network map comprises:
performing road thinning on the skeleton node segment, to obtain the two-dimensional road network map composed of breaks and a polygonal line segment.

4. The method according to claim 1 , wherein the obtaining the plane point cloud based on the three-dimensional point cloud data comprises:
filtering the three-dimensional point cloud data to obtain a three-dimensional ground point cloud; and
setting a z coordinate in the three-dimensional ground point cloud to be zero to obtain the plane point cloud.

5. The method according to claim 1 , wherein the determining, for the point in the two-dimensional road network map, the one or more associated points of the point in the plane point cloud, and determining the height of the point according to the elevation of the one or more associated points, to obtain the three-dimensional road network map comprises:
setting the point in the two-dimensional road network map as a current node;
performing a neighbor search on the plane point cloud with the current node as a center to obtain a plurality of associated points of the current node; and
determining a height of the current node according to elevations of the plurality of associated points and a preset threshold, to obtain the three-dimensional road network map.

6. The method according to claim 5 , wherein the determining the height of the current node according to the elevations of the plurality of associated points and the preset threshold comprises:
averaging the elevations of the plurality of associated points to obtain an average value;
calculating an elevation difference between each two associated points of the plurality of associated points;
comparing the elevation differences to determine a maximum elevation difference;
determining the average value as the height of the current node based on the maximum elevation difference being less than the preset threshold; and
determining a height of an associated point of the plurality of associated points closest to the current point as the height of the current node based on the maximum elevation difference being equal to or greater than the preset threshold.

7. The method according to claim 1 , further comprising:
constructing, for each road in the three-dimensional road network map, a vertical curve according to a distance from each point of the road to a starting point of the road and a height of each point of the road; and
performing smoothing on the vertical curve to update the height of each point of the road.

8. The method according to claim 1 , wherein the extracting the two-dimensional road based on the satellite remote sensing image comprises:
performing semantic segmentation for a road network in the satellite remote sensing image; and
converting the satellite remote sensing image after performing the semantic segmentation into a binary image to obtain the two-dimensional road.

9. An electronic device, comprising:
at least one processor; and
a memory communicatively connected with the at least one processor, wherein the memory stores instructions executable by the at least one processor that, when executed by the at least one processor, enable the at least one processor to execute actions including:
extracting a two-dimensional road based on a satellite remote sensing image;
skeletonizing the two-dimensional road to obtain a two-dimensional road network map; and
determining a height of each point in the two-dimensional road network map based on an elevation of three-dimensional point cloud data, to obtain the three-dimensional road network map, wherein the determining the height of each point in the two-dimensional road network map based on the elevation of the three-dimensional point cloud data, to obtain the three-dimensional road network map comprises:
obtaining a plane point cloud based on the three-dimensional point cloud data;
determining, for a point in the two-dimensional road network map, one or more associated points of the point in the plane point cloud; and
determining a height of the point according to elevation of the one or more associated points, to obtain the three-dimensional road network map.

10. The electronic device according to claim 9 , wherein the skeletonizing the two-dimensional road to obtain the two-dimensional road network map comprises:
skeletonizing the two-dimensional road, and determining a skeleton node segment based on skeleton nodes in the skeletonized two-dimensional road; and
converting, for the skeleton node segment, a pixel coordinate into a geographic coordinate, to obtain the two-dimensional road network map.

11. The electronic device according to claim 10 , wherein the converting, for the skeleton node segment, the pixel coordinate into the geographic coordinate, to obtain the two-dimensional road network map comprises:
performing road thinning on the skeleton node segment, to obtain the two-dimensional road network map composed of breaks and a polygonal line segment.

12. The electronic device according to claim 9 , wherein the obtaining the plane point cloud based on the three-dimensional point cloud data comprises:
filtering the three-dimensional point cloud data to obtain a three-dimensional ground point cloud; and
setting a z coordinate in the three-dimensional ground point cloud to be zero to obtain the plane point cloud.

13. The electronic device according to claim 9 , wherein the determining, for the point in the two-dimensional road network map, the one or more associated points of the point in the plane point cloud, and determining the height of the point according to the elevation of the one or more associated points, to obtain the three-dimensional road network map comprises:
setting the point in the two-dimensional road network map as a current node;
performing a neighbor search on the plane point cloud with the current node as a center to obtain a plurality of associated points of the current node; and
determining a height of the current node accord

Assignees

• Beijing Baidu Netcom Sci & Tech Co Ltd

Inventors

• Zhang Junjun

Classifications

• G06T2210/56

  Particle system, point based geometry or rendering · CPC title

• G06T2207/30181

  Earth observation · CPC title

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title

• G06T2207/10032

  Satellite or aerial image; Remote sensing · CPC title

• G06T2207/10028

  Range image; Depth image; 3D point clouds · CPC title