Enhancing Hyperspectral Images via Diffusion Model and Group-Autoencoder Super-resolution Network

• DOI: 10.1609/aaai.v38i6.28392
• Publication Date: 2024-03-25T09:44:51Z
• Authors: Zhaoyang Wang; Dongyang Li; Mingyang Zhang; Hao Luo; Maoguo Gong
• Keywords: Autoencoder; Diffusion imaging

Existing hyperspectral image (HSI) super-resolution (SR) methods struggle to effectively capture the complex spectral-spatial relationships and low-level details, while diffusion models represent a promising generative model known for their exceptional performance in modeling relations learning high visual features. The direct application of HSI SR is hampered by challenges such as difficulties convergence protracted inference time. In this work, we introduce novel Group-Autoencoder (GAE) framework that synergistically combines with construct highly effective (DMGASR). Our proposed GAE encodes high-dimensional data into low-dimensional latent space where works, thereby alleviating difficulty training maintaining band correlation considerably reducing Experimental results on both natural remote sensing datasets demonstrate method superior other state-of-the-art visually metrically.