Road network extraction method, device, and storage medium

Patent metadata

Field	Value
Publication number	US-12327407-B2
Application number	US-202318152119-A
Country	US
Kind code	B2
Filing date	Jan 9, 2023
Priority date	May 16, 2022
Publication date	Jun 10, 2025
Grant date	Jun 10, 2025

Abstract

Provided are a road network extraction method, a device, and a storage medium, which relate to the technical field of artificial intelligence and, in particular, to the fields of image processing, computer vision, and the like and are specifically applicable to scenarios such as intelligent transportation and a smart city. A specific implementation scheme includes: extracting a first road network of a target region according to user trajectories of the target region; extracting a second road network of the target region according to a satellite aerial image of the target region; and extract a target road network of the target region according to the first road network, the second road network, and the user trajectories. Efficient and accurate road network extraction can be achieved through techniques in embodiments of the present disclosure.

First claim

What is claimed is:

1. A road network extraction method, comprising:
extracting a first road network of a target region according to user trajectories of the target region;
extracting a second road network of the target region according to a satellite aerial image of the target region; and
extracting a target road network of the target region according to the first road network, the second road network, and the user trajectories;
wherein extracting the second road network of the target region according to the satellite aerial image of the target region comprises:
determining a feature map of the satellite aerial image of the target region;
determining, according to the feature map, a first probability of an optional pixel in the satellite aerial image being a road keypoint; and
extracting the second road network of the target region from the satellite aerial image according to the first probability;
wherein extracting the second road network of the target region from the satellite aerial image according to the first probability comprises:
selecting a target pixel from optional pixels according to the first probability;
extracting a region of interest of the target pixel from the feature map;
encoding the region of interest of the target pixel to obtain a local feature of the target pixel; and
extracting the second road network of the target region from the satellite aerial image according to a global feature of the target pixel in the feature map and the local feature of the target pixel;
wherein extracting the second road network of the target region from the satellite aerial image according to the global feature of the target pixel in the feature map and the local feature of the target pixel comprises:
determining a fused feature of the target pixel according to the global feature of the target pixel in the feature map and the local feature of the target pixel;
encoding the fused feature of the target pixel to obtain a second probability of the target pixel being the road keypoint, a third probability of the target pixel belonging to a road, and map encoding information of the target pixel; and
extracting the second road network of the target region from the satellite aerial image according to the second probability of the target pixel being the road keypoint, the third probability of the target pixel belonging to the road, and the map encoding information of the target pixel.

2. The method of claim 1 , wherein extracting the first road network of the target region according to the user trajectories of the target region comprises:
constructing a regional trajectory image according to the user trajectories of the target region;
determining a road network extraction strategy according to region heat of the target region; and
extracting the first road network of the target region from the regional trajectory image according to the road network extraction strategy.

3. The method of claim 2 , wherein extracting the first road network of the target region from the regional trajectory image according to the road network extraction strategy comprises:
in a case where the region heat is greater than a set heat threshold, determining a trajectory density peak profile of a road in the regional trajectory image according to the user trajectories; and
extracting the first road network of the target region from the regional trajectory image according to the trajectory density peak profile.

4. The method of claim 2 , wherein extracting the first road network of the target region from the regional trajectory image according to the road network extraction strategy comprises:
filtering the regional trajectory image in a filtering manner in the road network extraction strategy to obtain a filtered trajectory image; and
extracting the first road network of the target region from the filtered trajectory image according to a road network extraction manner in the road network extraction strategy.

5. The method of claim 1 , wherein extracting the target road network of the target region according to the first road network, the second road network, and the user trajectories comprises:
performing an intersection operation on the first road network and the second road network to obtain a public road network;
constructing a regional trajectory image according to the user trajectories of the target region;
extracting a third road network of the target region according to the second road network, the public road network, and the regional trajectory image; and
adding the third road network to the public road network to obtain the target road network of the target region.

6. The method of claim 5 , wherein performing the intersection operation on the first road network and the second road network to obtain the public road network comprises:
performing the intersection operation on the first road network and the second road network to obtain an intermediate road network; and
performing deduplication on the intermediate road network according to a road similarity to obtain the public road network.

7. The method of claim 5 , wherein extracting the third road network of the target region according to the second road network, the public road network, and the regional trajectory image comprises:
determining another road network in the second road network except the public road network; and
extracting the third road network from the another road network according to the regional trajectory image.

8. The method of claim 1 , further comprising:
determining a cumulative time length according to a total trajectory length in the target region; and
acquiring the user trajectories of the target region within the cumulative time length.

9. An electronic device, comprising:
at least one processor; and
a memory communicatively connected to the at least one processor;
wherein the memory stores an instruction executable by the at least one processor to enable the at least one processor to perform:
extracting a first road network of a target region according to user trajectories of the target region;
extracting a second road network of the target region according to a satellite aerial image of the target region; and
extracting a target road network of the target region according to the first road network, the second road network, and the user trajectories;
wherein the at least one processor extracts the second road network of the target region according to the satellite aerial image of the target region by:
determining a feature map of the satellite aerial image of the target region;
determining, according to the feature map, a first probability of an optional pixel in the satellite aerial image being a road keypoint; and extracting the second road network of the target region from the satellite aerial image according to the first probability;
wherein the at least one processor extracts the second road network of the target region from the satellite aerial image according to the first probability by:
selecting a target pixel from optional pixels according to the first probability;
extracting a region of interest of the target pixel from the feature map;
encoding the region of interest of the target pixel to obtain a local feature of the target pixel; and
extracting the second road network of the target region from the satellite aerial image according to a global feature of the target pixel in the feature map and the local feature of the target pixel; and
wherein the at least one processor extracts the second road network of the target region from the satellite aerial image according to the global feature of the target pixel in the feature map and the local feature of the target pixel by:
determining a fused feature o

Assignees

• Beijing Baidu Netcom Sci & Tech Co Ltd

Inventors

• Ye Xiaoqing
• Xia Deguo
• Huang Jizhou
• Wang Haifeng

Classifications

• G06V20/13

  Satellite images · CPC title

• G06V10/44

  Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components · CPC title

• G06V10/42

  Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation · CPC title

• G06V10/7715

  Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods · CPC title

• G06V10/25

  Determination of region of interest [ROI] or a volume of interest [VOI] · CPC title