Nonlinear spectral unmixing using residual component analysis and a Gamma Markov random field

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper presents a new Bayesian nonlinear unmixing model for hyperspectral images. The proposed model represents pixel reflectances as linear mixtures of endmembers, corrupted by an additional combination of nonlinear terms (with respect to the endmembers) and additive Gaussian noise. A central contribution of this work is to use a Gamma Markov random field to capture the spatial structure and correlations of the nonlinear terms, and by doing so to improve significantly estimation performance. In order to perform hyperspectral image unmixing, the Gamma Markov random field is embedded in a hierarchical Bayesian model representing the image observation process and prior knowledge, followed by inference with a Markov chain Monte Carlo algorithm that jointly estimates the model parameters of interest and marginalises latent variables. Simulations conducted with synthetic and real data show the accuracy of the proposed SU and nonlinearity estimation strategy for the analysis of hyperspectral images.

Original languageEnglish
Title of host publication2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP)
PublisherIEEE
Pages165-168
Number of pages4
ISBN (Print)9781479919635
DOIs
Publication statusPublished - 2015
Event6th IEEE International Workshop on Computational Advances in Multi-Sensor Adaptive Processing 2015 - Cancun, Mexico
Duration: 13 Dec 201516 Dec 2015

Conference

Conference6th IEEE International Workshop on Computational Advances in Multi-Sensor Adaptive Processing 2015
Abbreviated titleCAMSAP 2015
CountryMexico
CityCancun
Period13/12/1516/12/15

Fingerprint

Spectral Unmixing
Hyperspectral Image
Random Field
Hierarchical Bayesian Model
Markov Chain Monte Carlo Algorithms
Gaussian Noise
Spatial Structure
Latent Variables
Spatial Correlation
Term
Reflectance
Prior Knowledge
Pixel
Model
Nonlinearity
Markov processes
Pixels
Estimate
Simulation

Keywords

  • Bayesian estimation
  • Gamma Markov random field
  • Hyperspectral imagery
  • nonlinear spectral unmixing
  • residual component analysis

ASJC Scopus subject areas

  • Signal Processing
  • Computational Mathematics

Cite this

Altmann, Y., Pereyra, M., & McLaughlin, S. (2015). Nonlinear spectral unmixing using residual component analysis and a Gamma Markov random field. In 2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP) (pp. 165-168). IEEE. https://doi.org/10.1109/CAMSAP.2015.7383762
Altmann, Yoann ; Pereyra, Marcelo ; McLaughlin, Stephen. / Nonlinear spectral unmixing using residual component analysis and a Gamma Markov random field. 2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP). IEEE, 2015. pp. 165-168
@inproceedings{f9b538b70d684e5aa3880837cb550e0f,
title = "Nonlinear spectral unmixing using residual component analysis and a Gamma Markov random field",
abstract = "This paper presents a new Bayesian nonlinear unmixing model for hyperspectral images. The proposed model represents pixel reflectances as linear mixtures of endmembers, corrupted by an additional combination of nonlinear terms (with respect to the endmembers) and additive Gaussian noise. A central contribution of this work is to use a Gamma Markov random field to capture the spatial structure and correlations of the nonlinear terms, and by doing so to improve significantly estimation performance. In order to perform hyperspectral image unmixing, the Gamma Markov random field is embedded in a hierarchical Bayesian model representing the image observation process and prior knowledge, followed by inference with a Markov chain Monte Carlo algorithm that jointly estimates the model parameters of interest and marginalises latent variables. Simulations conducted with synthetic and real data show the accuracy of the proposed SU and nonlinearity estimation strategy for the analysis of hyperspectral images.",
keywords = "Bayesian estimation, Gamma Markov random field, Hyperspectral imagery, nonlinear spectral unmixing, residual component analysis",
author = "Yoann Altmann and Marcelo Pereyra and Stephen McLaughlin",
year = "2015",
doi = "10.1109/CAMSAP.2015.7383762",
language = "English",
isbn = "9781479919635",
pages = "165--168",
booktitle = "2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP)",
publisher = "IEEE",
address = "United States",

}

Altmann, Y, Pereyra, M & McLaughlin, S 2015, Nonlinear spectral unmixing using residual component analysis and a Gamma Markov random field. in 2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP). IEEE, pp. 165-168, 6th IEEE International Workshop on Computational Advances in Multi-Sensor Adaptive Processing 2015, Cancun, Mexico, 13/12/15. https://doi.org/10.1109/CAMSAP.2015.7383762

Nonlinear spectral unmixing using residual component analysis and a Gamma Markov random field. / Altmann, Yoann; Pereyra, Marcelo; McLaughlin, Stephen.

2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP). IEEE, 2015. p. 165-168.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Nonlinear spectral unmixing using residual component analysis and a Gamma Markov random field

AU - Altmann, Yoann

AU - Pereyra, Marcelo

AU - McLaughlin, Stephen

PY - 2015

Y1 - 2015

N2 - This paper presents a new Bayesian nonlinear unmixing model for hyperspectral images. The proposed model represents pixel reflectances as linear mixtures of endmembers, corrupted by an additional combination of nonlinear terms (with respect to the endmembers) and additive Gaussian noise. A central contribution of this work is to use a Gamma Markov random field to capture the spatial structure and correlations of the nonlinear terms, and by doing so to improve significantly estimation performance. In order to perform hyperspectral image unmixing, the Gamma Markov random field is embedded in a hierarchical Bayesian model representing the image observation process and prior knowledge, followed by inference with a Markov chain Monte Carlo algorithm that jointly estimates the model parameters of interest and marginalises latent variables. Simulations conducted with synthetic and real data show the accuracy of the proposed SU and nonlinearity estimation strategy for the analysis of hyperspectral images.

AB - This paper presents a new Bayesian nonlinear unmixing model for hyperspectral images. The proposed model represents pixel reflectances as linear mixtures of endmembers, corrupted by an additional combination of nonlinear terms (with respect to the endmembers) and additive Gaussian noise. A central contribution of this work is to use a Gamma Markov random field to capture the spatial structure and correlations of the nonlinear terms, and by doing so to improve significantly estimation performance. In order to perform hyperspectral image unmixing, the Gamma Markov random field is embedded in a hierarchical Bayesian model representing the image observation process and prior knowledge, followed by inference with a Markov chain Monte Carlo algorithm that jointly estimates the model parameters of interest and marginalises latent variables. Simulations conducted with synthetic and real data show the accuracy of the proposed SU and nonlinearity estimation strategy for the analysis of hyperspectral images.

KW - Bayesian estimation

KW - Gamma Markov random field

KW - Hyperspectral imagery

KW - nonlinear spectral unmixing

KW - residual component analysis

UR - http://www.scopus.com/inward/record.url?scp=84963815685&partnerID=8YFLogxK

U2 - 10.1109/CAMSAP.2015.7383762

DO - 10.1109/CAMSAP.2015.7383762

M3 - Conference contribution

SN - 9781479919635

SP - 165

EP - 168

BT - 2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP)

PB - IEEE

ER -

Altmann Y, Pereyra M, McLaughlin S. Nonlinear spectral unmixing using residual component analysis and a Gamma Markov random field. In 2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP). IEEE. 2015. p. 165-168 https://doi.org/10.1109/CAMSAP.2015.7383762