Physics-Guided Deep Unfolding for Blind Cross-Sensor Spectral Super-Resolution via Learning the Spectral Transformation Function
Researchers have developed a new physics-guided deep unfolding network called PGU-Net to tackle blind cross-sensor spectral super-resolution. This method can reconstruct hyperspectral images from multispectral images even when the spectral response function is unknown. PGU-Net jointly estimates the hyperspectral image and a learnable spectral transformation function, demonstrating improved reconstruction performance on benchmark datasets and real-world UAV data. AI
IMPACT This method could enable more cost-effective hyperspectral imaging by improving reconstruction from multispectral data.