Hyperspectral Band Selection based on Generalized 3DTV and Tensor CUR Decomposition
Katherine Henneberger, Jing Qin
TL;DR
This work tackles unsupervised hyperspectral band selection by modeling the data as a sum of a low-rank, spatial-spectral smooth tensor $B$ and a sparse tensor $S$, regularized with generalized 3D total variation ($G3DTV$) and solved via ADMM. A tensor-CUR decomposition update is used to efficiently enforce the low-rank constraint without resorting to costly full t-SVD, preserving the tensor structure of the data. The method demonstrates superior overall accuracy on Indian Pines and Salinas-A compared to several state-of-the-art baselines, particularly for larger numbers of selected bands, and provides practical parameter guidelines and noise-robustness insights. This tensor-based approach offers scalable, high-fidelity band selection with potential applicability to large-scale remote sensing datasets.
Abstract
Hyperspectral Imaging (HSI) serves as an important technique in remote sensing. However, high dimensionality and data volume typically pose significant computational challenges. Band selection is essential for reducing spectral redundancy in hyperspectral imagery while retaining intrinsic critical information. In this work, we propose a novel hyperspectral band selection model by decomposing the data into a low-rank and smooth component and a sparse one. In particular, we develop a generalized 3D total variation (G3DTV) by applying the $\ell_1^p$-norm to derivatives to preserve spatial-spectral smoothness. By employing the alternating direction method of multipliers (ADMM), we derive an efficient algorithm, where the tensor low-rankness is implied by the tensor CUR decomposition. We demonstrate the effectiveness of the proposed approach through comparisons with various other state-of-the-art band selection techniques using two benchmark real-world datasets. In addition, we provide practical guidelines for parameter selection in both noise-free and noisy scenarios.