.png){kind=icon}
.png){kind=icon}
Return
EB-Net: Adaptive Feature Sharpening for Open-Pit Mine Road Extraction From Remote Sensing Imagery
.png){kind=icon}
PRE
AI
Z
X
X
P
DOI：10.1109/lgrs.2026.3711867.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Accurate road extraction is critical for intelligent open-pit mine management; however, complex backgrounds, ambiguous boundaries, and irregular geometries pose significant challenges to existing methods. To address these issues, we propose EB-Net, a unified adaptive feature-sharpening network for open-pit mine road extraction. The proposed network integrates three complementary modules: a multiscale directional sharpening fusion (MDSF) module for enhanced spectral–spatial discrimination, an adaptive boundary refinement module (ABRM) for precise edge enhancement, and a road deformable block (RDBlock) for modeling geometric continuity. On the open-pit mine road (OPM) dataset, EB-Net outperforms D-SegNeXt by 1.52% IoU. On the public DeepGlobe benchmark, it attains 70.69% IoU, demonstrating strong generalization. These results indicate that EB-Net is a robust and effective solution for road extraction in complex open-pit mine scenarios and other remote sensing environments.
Keywords:
Boundary refinement
deformable convolution
feature sharpening
remote sensing images
and road extraction
Journal
I
IF:
4.4
Papers:
486
Citations:
0
Organization
Cited Papers