AI RESEARCH
Synthetic Aperture Radar Image Change Detection Based on Global Dynamic Context-Aware Network
arXiv CS.CV
•
ArXi:2605.16764v1 Announce Type: new Convolutional neural networks (CNNs) have been extensively and successfully applied to the task of synthetic aperture radar (SAR) image change detection. However, conventional convolutional layers are inherently limited by their local receptive fields, which mainly capture spatially localized patterns while neglecting the global context that is often crucial for accurately distinguishing subtle or large-scale changes in SAR imagery.