177 / 2023-11-27 14:11:31

Research on Denoising Methods for Hyperspectral Images Based on Low-Rank Theory and Sparse Representation

Hyperspectral Images, Noise Estimation, Low-Rank, Sparse Representation

This study addresses the issue of noise interference in hyperspectral images (HSI). By combining singular value decomposition (SVD) with an adaptive block algorithm, an improved algorithm for estimating noise intensity is proposed, aiming for precise assessment of noise levels. Additionally, an enhanced denoising method for hyperspectral images is introduced by integrating low-rank theory and sparse representation algorithms. The research results indicate that, for the Indian Pines public dataset, the denoising performance of the study surpasses existing algorithms by over 3.0 dB. Furthermore, robustness in estimating noise intensity is observed. Valuable insights for denoising similarly structured data with low signal-to-noise ratios are provided by this research, contributing meaningfully to the field.