Answers are grounded in this paper and cite their sources.
Heterogeneous remote sensing change detection focuses on analyzing multitemporal images captured by distinct sensors such as optical and SAR systems. This technique serves as an indispensable tool in fire surveillance, urban governance, and disaster emergency evaluation. However, it faces two core challenges: the lack of high-quality annotated datasets for complex fire scenarios, and severe cross-modal feature misalignment caused by inconsistent imaging mechanisms. To address these issues, this article first constructs the WFCD dataset, which integrates multitemporal Sentinel-1 SAR and Sentinel-2 optical images and provides pixel-level annotations to distinguish fire-affected areas. Second, we propose the Snake-Conv Mamba and feature alignment network (SCM-FANet). By integrating a dedicated feature alignment module to reduce cross-modal discrepancies and a snake-convolutional Mamba backbone for robust multiscale feature extraction, SCM-FANet effectively mitigates cross-modal feature misalignment and enhances the mining of change-related features, thereby improving detection accuracy and robustness. Extensive experiments are conducted on three representative datasets (WFCD, HWuhan, OMBRIA) covering complex fire, urban land cover, and flood scenarios. The results demonstrate that SCM-FANet consistently outperforms eight state-of-the-art methods in terms of F1-Score and IoU. The proposed dataset and method provide valuable support for heterogeneous remote sensing change detection in complex real-world scenarios.
No takes yet. Share an insight, caveat, or question.
Xu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: