Hyperspectral Image Superresolution Using Global Gradient Sparse and Nonlocal Low-Rank Tensor Decomposition With Hyper-Laplacian Prior
This article presents a novel global gradient sparse and nonlocal low-rank tensor decomposition model with a hyper-Laplacian prior for hyperspectral image (HSI) superresolution to produce a high-resolution HSI (HR-HSI) by fusing a low-resolution HSI (LR-HSI) with an HR multispectral image (HR-MSI)....
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2021-01-01
|
Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9417623/ |
id |
doaj-dc13c3fa6ba04d4596983c875e38e97c |
---|---|
record_format |
Article |
spelling |
doaj-dc13c3fa6ba04d4596983c875e38e97c2021-06-10T23:00:13ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352021-01-01145453546910.1109/JSTARS.2021.30761709417623Hyperspectral Image Superresolution Using Global Gradient Sparse and Nonlocal Low-Rank Tensor Decomposition With Hyper-Laplacian PriorYidong Peng0https://orcid.org/0000-0003-3779-0360Weisheng Li1https://orcid.org/0000-0002-9033-8245Xiaobo Luo2https://orcid.org/0000-0001-5688-0324Jiao Du3https://orcid.org/0000-0001-6402-1335Chongqing Key Laboratory of Image Cognition, Chongqing University of Posts and Telecommunications, Chongqing, ChinaChongqing Key Laboratory of Image Cognition, Chongqing University of Posts and Telecommunications, Chongqing, ChinaChongqing Engineering Research Center for Spatial Big Data Intelligent Technology, Chongqing Institute of Meteorological Science, Chongqing, ChinaSchool of Computer Science, and Cyber Engineering, Guangzhou University, Guangzhou, ChinaThis article presents a novel global gradient sparse and nonlocal low-rank tensor decomposition model with a hyper-Laplacian prior for hyperspectral image (HSI) superresolution to produce a high-resolution HSI (HR-HSI) by fusing a low-resolution HSI (LR-HSI) with an HR multispectral image (HR-MSI). Inspired by the investigated hyper-Laplacian distribution of the gradients of the difference images between the upsampled LR-HSI and latent HR-HSI, we formulate the relationship between these two datasets as a <inline-formula><tex-math notation="LaTeX">$\ell _{p}$</tex-math></inline-formula> <inline-formula><tex-math notation="LaTeX">$(0 < p < 1)$</tex-math></inline-formula>-norm term to enforce spectral preservation. Then, the relationship between the HR-MSI and latent HR-HSI is built using a tensor-based fidelity term to recover the spatial details. To effectively capture the high spatio-spectral-nonlocal similarities of the latent HR-HSI, we design a novel nonlocal low-rank Tucker decomposition to model the 3-D regular tensors constructed from the grouped nonlocal similar HR-HSI cubes. The global spatial-spectral total variation regularization is then adopted to ensure the global spatial piecewise smoothness and spectral consistency of the reconstructed HR-HSI from nonlocal low-rank cubes. Finally, an alternating direction method of multipliers-based algorithm is designed to efficiently solve the optimization problem. Experiments on both the synthetic and real datasets collected by different sensors show the effectiveness of the proposed method, from visual and quantitative assessments.https://ieeexplore.ieee.org/document/9417623/Global gradient sparsehyper-Laplacianhyperspectral imagenonlocal low-ranksuperresolutiontotal variation |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Yidong Peng Weisheng Li Xiaobo Luo Jiao Du |
spellingShingle |
Yidong Peng Weisheng Li Xiaobo Luo Jiao Du Hyperspectral Image Superresolution Using Global Gradient Sparse and Nonlocal Low-Rank Tensor Decomposition With Hyper-Laplacian Prior IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Global gradient sparse hyper-Laplacian hyperspectral image nonlocal low-rank superresolution total variation |
author_facet |
Yidong Peng Weisheng Li Xiaobo Luo Jiao Du |
author_sort |
Yidong Peng |
title |
Hyperspectral Image Superresolution Using Global Gradient Sparse and Nonlocal Low-Rank Tensor Decomposition With Hyper-Laplacian Prior |
title_short |
Hyperspectral Image Superresolution Using Global Gradient Sparse and Nonlocal Low-Rank Tensor Decomposition With Hyper-Laplacian Prior |
title_full |
Hyperspectral Image Superresolution Using Global Gradient Sparse and Nonlocal Low-Rank Tensor Decomposition With Hyper-Laplacian Prior |
title_fullStr |
Hyperspectral Image Superresolution Using Global Gradient Sparse and Nonlocal Low-Rank Tensor Decomposition With Hyper-Laplacian Prior |
title_full_unstemmed |
Hyperspectral Image Superresolution Using Global Gradient Sparse and Nonlocal Low-Rank Tensor Decomposition With Hyper-Laplacian Prior |
title_sort |
hyperspectral image superresolution using global gradient sparse and nonlocal low-rank tensor decomposition with hyper-laplacian prior |
publisher |
IEEE |
series |
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
issn |
2151-1535 |
publishDate |
2021-01-01 |
description |
This article presents a novel global gradient sparse and nonlocal low-rank tensor decomposition model with a hyper-Laplacian prior for hyperspectral image (HSI) superresolution to produce a high-resolution HSI (HR-HSI) by fusing a low-resolution HSI (LR-HSI) with an HR multispectral image (HR-MSI). Inspired by the investigated hyper-Laplacian distribution of the gradients of the difference images between the upsampled LR-HSI and latent HR-HSI, we formulate the relationship between these two datasets as a <inline-formula><tex-math notation="LaTeX">$\ell _{p}$</tex-math></inline-formula> <inline-formula><tex-math notation="LaTeX">$(0 < p < 1)$</tex-math></inline-formula>-norm term to enforce spectral preservation. Then, the relationship between the HR-MSI and latent HR-HSI is built using a tensor-based fidelity term to recover the spatial details. To effectively capture the high spatio-spectral-nonlocal similarities of the latent HR-HSI, we design a novel nonlocal low-rank Tucker decomposition to model the 3-D regular tensors constructed from the grouped nonlocal similar HR-HSI cubes. The global spatial-spectral total variation regularization is then adopted to ensure the global spatial piecewise smoothness and spectral consistency of the reconstructed HR-HSI from nonlocal low-rank cubes. Finally, an alternating direction method of multipliers-based algorithm is designed to efficiently solve the optimization problem. Experiments on both the synthetic and real datasets collected by different sensors show the effectiveness of the proposed method, from visual and quantitative assessments. |
topic |
Global gradient sparse hyper-Laplacian hyperspectral image nonlocal low-rank superresolution total variation |
url |
https://ieeexplore.ieee.org/document/9417623/ |
work_keys_str_mv |
AT yidongpeng hyperspectralimagesuperresolutionusingglobalgradientsparseandnonlocallowranktensordecompositionwithhyperlaplacianprior AT weishengli hyperspectralimagesuperresolutionusingglobalgradientsparseandnonlocallowranktensordecompositionwithhyperlaplacianprior AT xiaoboluo hyperspectralimagesuperresolutionusingglobalgradientsparseandnonlocallowranktensordecompositionwithhyperlaplacianprior AT jiaodu hyperspectralimagesuperresolutionusingglobalgradientsparseandnonlocallowranktensordecompositionwithhyperlaplacianprior |
_version_ |
1721384429810614272 |