SAGN: Sharpening-Aware Graph Network for Hyperspectral Image Change Detection
Bing Yang,Weiwei Sun,Jiangtao Peng
DOI: https://doi.org/10.1109/tgrs.2024.3403971
IF: 8.2
2024-06-04
IEEE Transactions on Geoscience and Remote Sensing
Abstract:Graph neural networks (GNNs) have garnered significant attention in hyperspectral image (HSI) change detection (CD). However, existing GNN-based methods extract features by aggregating neighborhood information, which is essentially a low-pass Laplacian smoothing operation and tends to diminish change information between bitemporal HSIs. In addition, these methods rely on fixed hand-crafted graphs, and thus cannot capture complex structures of HSIs well. To address these deficiencies, this article develops a sharpening-aware graph network (SAGN) for achieving high-quality HSI CD. First, to counteract the weakening of differences caused by Laplacian smoothing, this article proposes a novel Laplacian sharpening-based graph convolution (LSGC) module to accentuate change information between bitemporal HSIs. Second, instead of using "similarity graphs," this article constructs untied "difference graphs" for bitemporal HSIs to model dissimilarities between changed pixels and their neighbors. The SAGN can dynamically update graph structures across all layers, aiming to further maximize the divergence. Finally, a joint loss function, incorporating modified cross-entropy loss and contrastive loss, is devised to enhance interclass discrimination of learned features and alleviate the issues stemming from imbalanced labeled samples. Experiments on various HSI CD datasets demonstrate the effectiveness and superiority of the proposed SAGN.
imaging science & photographic technology,remote sensing,engineering, electrical & electronic,geochemistry & geophysics