A geological key target detection algorithm in complex multisource satellite images

  • Yaoyao Li
  • , Jingchun Cheng*
  • , Weiqi Huang
  • , Fan Zhang
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Geological target detection can help people understand and utilize the natural environment by accurately recognizing surface elements like as mountains, rivers, vegetation distribution, and etc. Most current methods use multi-source satellite imagery for geological target detection. However, most algorithms simply stack multi-source images as direct input without performing preliminary information purification. In this paper, we propose that geological target detection can be enhanced with sensitivity to the landform and terrain information which is critical for recognition. To be specific, we design a simple scheme to extract the landform cue and terrain cue from multi-source satellite images, and use them as inputs to build a three-branch landform- and terrain-aware detection framework. We validate the effectiveness of each cue and the advancing performance of our proposed detection framework on the Landslide4Sense dataset via quantitative evaluation and qualitative analysis.

Original languageEnglish
Title of host publicationFourth International Conference on Image Processing, Object Detection, and Tracking, IPODT 2025
EditorsYang Zhao, Bao Feng, Lingxia Liao
PublisherSPIE
ISBN (Electronic)9781510699663
DOIs
StatePublished - 30 Dec 2025
Event4th International Conference on Image Processing, Object Detection, and Tracking, IPODT 2025 - Guilin, China
Duration: 19 Sep 202521 Sep 2025

Publication series

NameProceedings of SPIE - The International Society for Optical Engineering
Volume14005
ISSN (Print)0277-786X
ISSN (Electronic)1996-756X

Conference

Conference4th International Conference on Image Processing, Object Detection, and Tracking, IPODT 2025
Country/TerritoryChina
CityGuilin
Period19/09/2521/09/25

Keywords

  • geological target detection
  • landform cue
  • terrain cue
  • three-branch network

Fingerprint

Dive into the research topics of 'A geological key target detection algorithm in complex multisource satellite images'. Together they form a unique fingerprint.

Cite this