Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The authors formulate multispectral-to-hyperspectral super-resolution as an operator learning problem, implementing a Spatial-Spectral Residual Operator Network (SSRON) to map discrete Sentinel-2A multispectral inputs to continuous EMIT-like hyperspectral profiles. By learning function-to-function mappings rather than static vector projections, the model circumvents sensor-specific spectral discretization constraints. SSRON outperforms existing deep learning baselines across all metrics while demonstrating zero-shot interpolation to unseen bands and enabling spectral query evaluation at finer wavelength intervals than the target sensor's native grid.
Treating satellite spectra as continuous functions rather than discrete vectors unlocks zero-shot reconstruction of completely unseen spectral bands at resolutions finer than native sensors.
Spectral super-resolution of multispectral satellite images can enable high temporal- and spatial-resolution hyperspectral satellite imagery at a modest cost, significantly increasing the applicability of hyperspectral remote sensing. This task is inherently ill-posed, making it well-suited for deep learning-based methods. In this study, the spectral super-resolution task is framed as an operator learning problem, and SSRON is proposed as a Deep Operator Network that effectively learns function-to-function mappings from downsampled spectra to continuous spectra. The model is trained to super-resolve Sentinel-2A-like multispectral imagery to EMIT images. Compared to baseline models, SSRON achieves superior performance across all metrics. The model also demonstrates zero-shot spectral super-resolution capability by predicting bands unseen during training. Furthermore, its continuous-output formulation suggests the potential to estimate spectra at finer wavelength intervals than the native sensor. These results suggest the potential of SSRON and establishes operator learning as a promising direction for spectral super-resolution.