Spectral super-resolution via physics-guided deep unfolding

Arxiv CS.CV (computer vision)

Paper proposes a physics-guided deep unfolding method for blind cross-sensor spectral super-resolution, reconstructing hyperspectral images from RGB inputs. The approach learns a spectral transformation function to handle sensor differences, targeting remote sensing applications where dedicated hyperspectral sensors are unavailable.

deep-learning

Arxiv CS.CV (computer vision)·Zhaolin LiJinsong ChenShanxin GuoTuo ZhangXinglong ZhangPan Chen

6 days ago