Dual-Domain Dynamic Local–Global Network for Pansharpening

Pansharpening has benefited from the development of deep learning (DL) and has achieved excellent results. However, most DL-based methods extract local features by convolutional neural networks and do not integrate global features. Moreover, these methods only extract high-frequency features on the high-pass domain (HPD) or only consider image features on the intensity domain (ID). The method that only considers features in one domain may result in insufficient extraction of spatial and spectral features. Therefore, we propose a dynamic local-global network model on dual-domains, that is, HPD and ID. The dynamic local-global feature extraction block (DLGB) is designed to dynamically integrate local and global features to improve the representation capability of the network. To decrease the computational complexity of global feature extraction, a lightweight biaxial nonlocal attention (BNLA) that captures global spatial features in horizontal and vertical directions is proposed. Experiments on GeoEye-1, QuickBird, and WorldView-3 datasets show that the proposed method presents better fusion performance on objective evaluation indices and subjective perception.

Automated summary

Pansharpening has achieved excellent results with the development of deep learning, but most methods only extract local features and ignore global features. In this study, a dynamic local-global network model is proposed on dual-domains, integrating local and global features to improve network representation capability. Experiments demonstrate that the proposed method has better fusion performance in objective evaluation and subjective perception.