Low-Light Hyperspectral Image Enhancement

Due to inadequate energy captured by the hyperspectral camera sensor in poor illumination conditions, low-light images (HSIs) usually suffer from low visibility, spectral distortion, and various noises. A range of HSI restoration methods have been developed, yet their effectiveness enhancing HSIs is constrained. This work focuses on enhancement task, which aims reveal spatial-spectral information hidden darkened areas. To facilitate development processing, we collect a (LHSI) dataset both indoor outdoor scenes. Based Laplacian pyramid decomposition reconstruction, developed an end-to-end data-driven (HSIE) approach trained LHSI dataset. With observation that related low-frequency component HSI, while textural details are closely correlated high-frequency component, proposed HSIE designed two branches. The branch adopted enlighten with reduced resolution. refinement utilized for refining via predicted mask. In addition, improve flow boost performance, introduce effective channel attention block (CAB) residual dense connection, served as basic branch. efficiency quantitative assessment measures visual effects demonstrated experimental results According classification performance remote sensing Indian Pines dataset, downstream tasks benefit enhanced HSI. Datasets codes available: https://github.com/guanguanboy/HSIE.