Zero-shot Hyperspectral Image Denoising using self-completion with 3D Random patterned masks

Hyperspectral images (HSIs) have higher spectral resolution than RGB and are used in various tasks. However, HSIs prone to degradation due noise generated during imaging, making it difficult obtain non-degraded images. Additionally, supervised learning, which relies on pairs of degraded images, is often challenging apply HSI restoration because the high cost imaging need prepare large amounts data. To overcome these limitations, recent advances self-supervised learning led development learning-based image methods that do not require including low accuracy estimate distribution. In this paper, we propose a zero-shot deep denoising method based restoration. The proposed achieves recovery by repeatedly predicting blind-spots 3D blocks process. Notably, our does training or clean nor rely distribution information. Numerical experiments ablation studies confirmed comparable better conventional methods.