Hyperspectral Image Denoising

Hyperspectral image denoising aims to remove noise from hyperspectral images, improving their quality for various applications. Current research heavily focuses on leveraging both local and non-local spatial-spectral correlations using advanced model architectures, including transformers, recurrent neural networks, and diffusion models, often combined with convolutional neural networks for efficient feature extraction. These methods aim to improve denoising performance while maintaining computational efficiency, addressing the challenges posed by the high dimensionality of hyperspectral data. The resulting improvements in image quality have significant implications for diverse fields like remote sensing, precision agriculture, and medical imaging.