Nonnegative Matrix Factorization With Data-Guided Constraints For Hyperspectral Unmixing
Hyperspectral unmixing aims to estimate a set of endmembers and corresponding abundances in pixels. Nonnegative matrix factorization (NMF) and its extensions with various constraints have been widely applied to hyperspectral unmixing. L 1 / 2 and L 2 regularizers can be added to NM...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2017-10-01
|
Series: | Remote Sensing |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-4292/9/10/1074 |
id |
doaj-262f9463254848bd91a241eec0cc7a74 |
---|---|
record_format |
Article |
spelling |
doaj-262f9463254848bd91a241eec0cc7a742020-11-25T01:49:57ZengMDPI AGRemote Sensing2072-42922017-10-01910107410.3390/rs9101074rs9101074Nonnegative Matrix Factorization With Data-Guided Constraints For Hyperspectral UnmixingRisheng Huang0Xiaorun Li1Liaoying Zhao2College of Electrical Engineering, Zhejiang University, No.38, Zheda Road, Xihu District, Hangzhou 310027, ChinaCollege of Electrical Engineering, Zhejiang University, No.38, Zheda Road, Xihu District, Hangzhou 310027, ChinaSchool of Computer Science, Hangzhou Dianzi University, Hangzhou 310018, ChinaHyperspectral unmixing aims to estimate a set of endmembers and corresponding abundances in pixels. Nonnegative matrix factorization (NMF) and its extensions with various constraints have been widely applied to hyperspectral unmixing. L 1 / 2 and L 2 regularizers can be added to NMF to enforce sparseness and evenness, respectively. In practice, a region in a hyperspectral image may possess different sparsity levels across locations. The problem remains as to how to impose constraints accordingly when the level of sparsity varies. We propose a novel nonnegative matrix factorization with data-guided constraints (DGC-NMF). The DGC-NMF imposes on the unknown abundance vector of each pixel with either an L 1 / 2 constraint or an L 2 constraint according to its estimated mixture level. Experiments on the synthetic data and real hyperspectral data validate the proposed algorithm.https://www.mdpi.com/2072-4292/9/10/1074nonnegative matrix factorizationdata-guided constraintssparsenessevenness |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Risheng Huang Xiaorun Li Liaoying Zhao |
spellingShingle |
Risheng Huang Xiaorun Li Liaoying Zhao Nonnegative Matrix Factorization With Data-Guided Constraints For Hyperspectral Unmixing Remote Sensing nonnegative matrix factorization data-guided constraints sparseness evenness |
author_facet |
Risheng Huang Xiaorun Li Liaoying Zhao |
author_sort |
Risheng Huang |
title |
Nonnegative Matrix Factorization With Data-Guided Constraints For Hyperspectral Unmixing |
title_short |
Nonnegative Matrix Factorization With Data-Guided Constraints For Hyperspectral Unmixing |
title_full |
Nonnegative Matrix Factorization With Data-Guided Constraints For Hyperspectral Unmixing |
title_fullStr |
Nonnegative Matrix Factorization With Data-Guided Constraints For Hyperspectral Unmixing |
title_full_unstemmed |
Nonnegative Matrix Factorization With Data-Guided Constraints For Hyperspectral Unmixing |
title_sort |
nonnegative matrix factorization with data-guided constraints for hyperspectral unmixing |
publisher |
MDPI AG |
series |
Remote Sensing |
issn |
2072-4292 |
publishDate |
2017-10-01 |
description |
Hyperspectral unmixing aims to estimate a set of endmembers and corresponding abundances in pixels. Nonnegative matrix factorization (NMF) and its extensions with various constraints have been widely applied to hyperspectral unmixing. L 1 / 2 and L 2 regularizers can be added to NMF to enforce sparseness and evenness, respectively. In practice, a region in a hyperspectral image may possess different sparsity levels across locations. The problem remains as to how to impose constraints accordingly when the level of sparsity varies. We propose a novel nonnegative matrix factorization with data-guided constraints (DGC-NMF). The DGC-NMF imposes on the unknown abundance vector of each pixel with either an L 1 / 2 constraint or an L 2 constraint according to its estimated mixture level. Experiments on the synthetic data and real hyperspectral data validate the proposed algorithm. |
topic |
nonnegative matrix factorization data-guided constraints sparseness evenness |
url |
https://www.mdpi.com/2072-4292/9/10/1074 |
work_keys_str_mv |
AT rishenghuang nonnegativematrixfactorizationwithdataguidedconstraintsforhyperspectralunmixing AT xiaorunli nonnegativematrixfactorizationwithdataguidedconstraintsforhyperspectralunmixing AT liaoyingzhao nonnegativematrixfactorizationwithdataguidedconstraintsforhyperspectralunmixing |
_version_ |
1725003786997465088 |