Abstract: Both traditional and learning-based hyperspectral image (HSI) compression methods suffer from significant quality loss at high compression ratios. To address this, we propose a low-overhead, ...