Spectral Unmixing-Guided Prompt Network for Cross-Domain Hyperspectral Image Classification

In the field of cross-domain hyperspectral image classification, single-source domain generalization (DG) frameworks are currently being explored. Most existing methods generate out-of-domain data from the source domain (SD) using techniques like style transfer, photometric, and geometric transformations. However, the domain expansion strategies designed by these methods often ignore the characteristics of mixed pixels in hyperspectral images. In addition, current hyperspectral imagery (HSI) cross-domain classification methods mainly focus on domain-invariant representations, neglecting the use of domain-specific knowledge for cooperative inference in the target domain (TD). To address these limitations, we propose the spectral unmixing-guided prompt network (SUPnet), which integrates unmixing and classification through a coordinated mechanism. This approach generates out-of-domain data based on spectral unmixing and employs a prompt learning strategy to facilitate multidomain-specific representation collaborative inference. Specifically, in order to estimate abundance values and endmembers in an unsupervised manner, a spectral unmixing generator (SUG) is designed, which constructs extended domains (EDs) by perturbing endmembers weighted by abundance values. Furthermore, a multilevel prompt learning (MPL) is developed to perform prompt learning at both the domain and endmember levels, thereby extracting specific knowledge and correlations between different domains to improve TD inference. Experimental results on five cross-domain HSI datasets covering multiple platforms (including UVA, airborne, and satellite) demonstrate that the proposed method significantly outperforms several current state-of-the-art approaches. Notably, SUG proves to be more efficient than many of advanced unmixing methods. The code will be available at: https://github.com/YuxiangZhang-BIT/IEEE_TGRS_SUPnet