Hyperspectral unmixing by reweighted low rank and total variation

Rui Wang, Wenzhi Liao, Heng-Chao Li, Hongyan Zhang, Aleksandra Pizurica

Research output: Contribution to conferencePaper

2 Citations (Scopus)

Abstract

In recent years, sparse regression has drawn much attention in hyperspectral unmixing. The well known sparse unmixing via variable splitting augmented Lagrangian (SUnSAL) and sparse unmixing via variable splitting augmented Lagrangian and total variation (SUnSAl-TV) aim to find the sparsest abundance of every data vector individually. However, these methods ignore the global structure of all the vectors. In this paper, we propose a novel hyperspectral unmixing method by exploiting low rank property of the abundance matrix. Our proposed method find the lowest-rank representation of a collection of the abundance vectors by using reweighted low rank constraint. This way, our proposed unmixing method better captures the global structure of the abundance matrix and improve the accuracy of abundance estimation. Our approach also takes the spatial context into account by a TV constraint. Experimental results on both the synthetic and real hyperspectral data demonstrate the effectiveness of our proposed algorithm.
Original languageEnglish
Number of pages4
DOIs
Publication statusPublished - 19 Oct 2017
Event2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS) - Los Angeles, United States
Duration: 21 Aug 201624 Aug 2016

Workshop

Workshop2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)
Abbreviated titleWHISPERS 2016
CountryUnited States
CityLos Angeles
Period21/08/1624/08/16

Fingerprint

abundance estimation
capture method
matrix
matrices
regression analysis
method

Keywords

  • unmixing
  • hyperspectral remote sensing
  • reweighted
  • low rank
  • hyperspectral Imaging
  • geophysical image processing
  • regression analysis
  • optimization

Cite this

Wang, R., Liao, W., Li, H-C., Zhang, H., & Pizurica, A. (2017). Hyperspectral unmixing by reweighted low rank and total variation. Paper presented at 2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Los Angeles, United States. https://doi.org/10.1109/WHISPERS.2016.8071668
Wang, Rui ; Liao, Wenzhi ; Li, Heng-Chao ; Zhang, Hongyan ; Pizurica, Aleksandra. / Hyperspectral unmixing by reweighted low rank and total variation. Paper presented at 2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Los Angeles, United States.4 p.
@conference{3a73d6088ce840738b9f450458827543,
title = "Hyperspectral unmixing by reweighted low rank and total variation",
abstract = "In recent years, sparse regression has drawn much attention in hyperspectral unmixing. The well known sparse unmixing via variable splitting augmented Lagrangian (SUnSAL) and sparse unmixing via variable splitting augmented Lagrangian and total variation (SUnSAl-TV) aim to find the sparsest abundance of every data vector individually. However, these methods ignore the global structure of all the vectors. In this paper, we propose a novel hyperspectral unmixing method by exploiting low rank property of the abundance matrix. Our proposed method find the lowest-rank representation of a collection of the abundance vectors by using reweighted low rank constraint. This way, our proposed unmixing method better captures the global structure of the abundance matrix and improve the accuracy of abundance estimation. Our approach also takes the spatial context into account by a TV constraint. Experimental results on both the synthetic and real hyperspectral data demonstrate the effectiveness of our proposed algorithm.",
keywords = "unmixing, hyperspectral remote sensing, reweighted, low rank, hyperspectral Imaging, geophysical image processing, regression analysis, optimization",
author = "Rui Wang and Wenzhi Liao and Heng-Chao Li and Hongyan Zhang and Aleksandra Pizurica",
year = "2017",
month = "10",
day = "19",
doi = "10.1109/WHISPERS.2016.8071668",
language = "English",
note = "2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), WHISPERS 2016 ; Conference date: 21-08-2016 Through 24-08-2016",

}

Wang, R, Liao, W, Li, H-C, Zhang, H & Pizurica, A 2017, 'Hyperspectral unmixing by reweighted low rank and total variation' Paper presented at 2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Los Angeles, United States, 21/08/16 - 24/08/16, . https://doi.org/10.1109/WHISPERS.2016.8071668

Hyperspectral unmixing by reweighted low rank and total variation. / Wang, Rui; Liao, Wenzhi; Li, Heng-Chao; Zhang, Hongyan; Pizurica, Aleksandra.

2017. Paper presented at 2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Los Angeles, United States.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Hyperspectral unmixing by reweighted low rank and total variation

AU - Wang, Rui

AU - Liao, Wenzhi

AU - Li, Heng-Chao

AU - Zhang, Hongyan

AU - Pizurica, Aleksandra

PY - 2017/10/19

Y1 - 2017/10/19

N2 - In recent years, sparse regression has drawn much attention in hyperspectral unmixing. The well known sparse unmixing via variable splitting augmented Lagrangian (SUnSAL) and sparse unmixing via variable splitting augmented Lagrangian and total variation (SUnSAl-TV) aim to find the sparsest abundance of every data vector individually. However, these methods ignore the global structure of all the vectors. In this paper, we propose a novel hyperspectral unmixing method by exploiting low rank property of the abundance matrix. Our proposed method find the lowest-rank representation of a collection of the abundance vectors by using reweighted low rank constraint. This way, our proposed unmixing method better captures the global structure of the abundance matrix and improve the accuracy of abundance estimation. Our approach also takes the spatial context into account by a TV constraint. Experimental results on both the synthetic and real hyperspectral data demonstrate the effectiveness of our proposed algorithm.

AB - In recent years, sparse regression has drawn much attention in hyperspectral unmixing. The well known sparse unmixing via variable splitting augmented Lagrangian (SUnSAL) and sparse unmixing via variable splitting augmented Lagrangian and total variation (SUnSAl-TV) aim to find the sparsest abundance of every data vector individually. However, these methods ignore the global structure of all the vectors. In this paper, we propose a novel hyperspectral unmixing method by exploiting low rank property of the abundance matrix. Our proposed method find the lowest-rank representation of a collection of the abundance vectors by using reweighted low rank constraint. This way, our proposed unmixing method better captures the global structure of the abundance matrix and improve the accuracy of abundance estimation. Our approach also takes the spatial context into account by a TV constraint. Experimental results on both the synthetic and real hyperspectral data demonstrate the effectiveness of our proposed algorithm.

KW - unmixing

KW - hyperspectral remote sensing

KW - reweighted

KW - low rank

KW - hyperspectral Imaging

KW - geophysical image processing

KW - regression analysis

KW - optimization

UR - http://hdl.handle.net/1854/LU-7221571

U2 - 10.1109/WHISPERS.2016.8071668

DO - 10.1109/WHISPERS.2016.8071668

M3 - Paper

ER -

Wang R, Liao W, Li H-C, Zhang H, Pizurica A. Hyperspectral unmixing by reweighted low rank and total variation. 2017. Paper presented at 2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Los Angeles, United States. https://doi.org/10.1109/WHISPERS.2016.8071668