A graph contrastive learning network for change detection with heterogeneous remote sensing images

Open Webpage

Zhiyong Lv, Sizhe Cheng, Linfu Xie, Junhuai Li, Minghua Zhao

Published: 01 Jan 2026, Last Modified: 04 Nov 2025Pattern Recognit. 2026EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•GCLN is proposed for unsupervised change detection with a novel contrastive loss.•AAP is first applied in preserving local connections and low-level spatial features.•Novel feature selection strategy is developed to remove abnormal vertices in graphs.