A mountainous road information automatic extraction method based on multi-source remote sensing data

By combining feature fusion and diffusion models of optical remote sensing imagery and SAR imagery, the problem of incomplete road information extraction in mountainous areas was solved, achieving efficient and accurate road extraction and providing reliable data support for disaster relief in mountainous areas.

CN122116117APending Publication Date: 2026-05-29INST OF GEOGRAPHIC SCI HEBEI ACAD OF SCI
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
INST OF GEOGRAPHIC SCI HEBEI ACAD OF SCI
Filing Date
2026-01-15
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing technologies often fail to extract complete road information in mountainous environments due to tree obstruction, affecting the accuracy and efficiency of disaster relief.

Method used

A method based on multi-source remote sensing data is adopted, combining optical remote sensing imagery and SAR imagery. Feature fusion and noise removal are performed through a diffusion model, and road extraction is carried out using a cross-attention module and a temporal embedding DCSwin model. A road extraction model is constructed to achieve end-to-end road segmentation.

Benefits of technology

It improves the accuracy and completeness of mountain road extraction, provides efficient and reliable data support for disaster relief, reduces dependence on large-scale labeled datasets, and has better generalization ability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122116117A_ABST
    Figure CN122116117A_ABST
Patent Text Reader

Abstract

The present application relates to the field of information extraction of remote sensing data, and discloses a mountainous area road information automatic extraction method based on multi-source remote sensing data, comprising: acquiring a multi-modal remote sensing image training sample set, the multi-modal remote sensing image training sample set comprising multiple groups of multi-modal remote sensing image training samples, each group of multi-modal remote sensing image training samples comprising a multi-modal remote sensing image and a road label image corresponding to the multi-modal remote sensing image; constructing an initial road extraction model, training the constructed initial model through the acquired multi-modal remote sensing image training sample set until the loss function converges to a preset value, obtaining a trained road extraction model; based on the trained road extraction model, performing road extraction on input multi-modal remote sensing images and vectorizing to obtain mountainous area road extraction information. The present application solves the problem of inaccurate and incomplete mountainous area road extraction in the prior art, greatly improving the accuracy and completeness of the extracted mountainous area road.
Need to check novelty before this filing date? Find Prior Art