Related Experiment Video
Updated: Feb 28, 2026
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
1.1K
EHDCD: An Edge Enhanced Hierarchical Dual Gated Network for Forest-Cropland Change Detection.
Tingting Zhao1, Yicong Sun1, Xia Yu1
1College of Computer and Information Engineering, Inner Mongolia Agricultural University, Hohhot 010018, China.
Sensors (Basel, Switzerland)
|February 27, 2026
Summary
A new Edge Enhanced Hierarchical Dual Gated Change Detection (EHDCD) model improves remote sensing change detection for forested and cultivated land. This method accurately identifies land cover changes, enhancing monitoring applications.
Area of Science:
- Remote Sensing
- Geospatial Analysis
- Environmental Monitoring
Background:
- Existing remote sensing change detection methods struggle with fine edge structures and distinguishing pseudo changes.
- Forested and cultivated land exhibit distinct spatial spectral attributes, posing challenges for accurate land cover classification.
- There is a need for advanced models to represent complex features and improve change detection accuracy for these land types.
Purpose of the Study:
- To introduce an Edge Enhanced Hierarchical Dual Gated Change Detection (EHDCD) model for improved change detection between forested and cultivated land.
- To enhance the representation of complex features specific to forested and cultivated land in remote sensing images.
- To address the limitations of current methods in capturing fine edge structures and reducing pseudo-change detection.
Main Methods:
- Development of an Edge Enhanced Channel Attention Module (EECA) to improve edge recognition and noise suppression.
- Implementation of a High-Low Level Dynamic Adaptation Strategy (HiLo) for balancing detail and semantic features.
- Construction of a Dual Gated Feature Compensation Module (DGFM) to minimize misdetection rates in change detection.
Main Results:
- The EHDCD model achieved F1 scores of 89.06% on the FC-CD dataset, 83.37% on CLCD, and 85.06% on SYSU-CD.
- The model demonstrated superior performance in capturing fine edge structures and distinguishing actual changes from noise.
- Experimental results confirm the model's effectiveness in accurately detecting changes between forested and cultivated land.
Conclusions:
- The EHDCD model offers a significant advancement in remote sensing change detection for forested and cultivated land.
- The proposed modules (EECA, HiLo, DGFM) effectively address the limitations of existing methods.
- The model provides more accurate support for dynamic monitoring applications of forest land and cropland.
