Supervised hyperspectral image segmentation

A convex formulation using hidden fields

Filipe Condessa, José Bioucas-Dias, Jelena Kovacevic

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

Abstract

Image segmentation is fundamentally a discrete problem. It consists of finding a partition of the image domain such that the pixels in each element of the partition exhibit some kind of similarity. The optimization is obtained via integer optimization which is NP-hard, apart from few exceptions. We sidestep from the discrete nature of image segmentation by formulating the problem in the Bayesian framework and introducing a hidden set of real-valued random fields determining the probability of a given partition. Armed with this model, the original discrete optimization is converted into a convex program. To infer the hidden fields, we introduce the Segmentation via the Constrained Split Augmented Lagrangian Shrinkage Algorithm (SegSALSA). The effectiveness of the proposed methodology is illustrated with hyperspectral image segmentation.

Original languageEnglish (US)
Title of host publication2014 6th Workshop on Hyperspectral Image and Signal Processing
Subtitle of host publicationEvolution in Remote Sensing, WHISPERS 2014
PublisherIEEE Computer Society
Volume2014-June
ISBN (Electronic)9781467390125
DOIs
StatePublished - Oct 19 2017
Event6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014 - Lausanne, Switzerland
Duration: Jun 24 2014Jun 27 2014

Other

Other6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014
CountrySwitzerland
CityLausanne
Period6/24/146/27/14

Fingerprint

Image segmentation
Pixels

Keywords

  • alternating optimization
  • Constrained Split Augmented Lagrangian Shrinkage Algorithm (SALSA)
  • hidden fields
  • hidden Markov measure fields
  • Image segmentation

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Signal Processing

Cite this

Condessa, F., Bioucas-Dias, J., & Kovacevic, J. (2017). Supervised hyperspectral image segmentation: A convex formulation using hidden fields. In 2014 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014 (Vol. 2014-June). [8077490] IEEE Computer Society. https://doi.org/10.1109/WHISPERS.2014.8077490

Supervised hyperspectral image segmentation : A convex formulation using hidden fields. / Condessa, Filipe; Bioucas-Dias, José; Kovacevic, Jelena.

2014 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014. Vol. 2014-June IEEE Computer Society, 2017. 8077490.

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

Condessa, F, Bioucas-Dias, J & Kovacevic, J 2017, Supervised hyperspectral image segmentation: A convex formulation using hidden fields. in 2014 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014. vol. 2014-June, 8077490, IEEE Computer Society, 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014, Lausanne, Switzerland, 6/24/14. https://doi.org/10.1109/WHISPERS.2014.8077490
Condessa F, Bioucas-Dias J, Kovacevic J. Supervised hyperspectral image segmentation: A convex formulation using hidden fields. In 2014 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014. Vol. 2014-June. IEEE Computer Society. 2017. 8077490 https://doi.org/10.1109/WHISPERS.2014.8077490
Condessa, Filipe ; Bioucas-Dias, José ; Kovacevic, Jelena. / Supervised hyperspectral image segmentation : A convex formulation using hidden fields. 2014 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014. Vol. 2014-June IEEE Computer Society, 2017.
@inproceedings{2f5381953e234e13803bcc15d2f13d38,
title = "Supervised hyperspectral image segmentation: A convex formulation using hidden fields",
abstract = "Image segmentation is fundamentally a discrete problem. It consists of finding a partition of the image domain such that the pixels in each element of the partition exhibit some kind of similarity. The optimization is obtained via integer optimization which is NP-hard, apart from few exceptions. We sidestep from the discrete nature of image segmentation by formulating the problem in the Bayesian framework and introducing a hidden set of real-valued random fields determining the probability of a given partition. Armed with this model, the original discrete optimization is converted into a convex program. To infer the hidden fields, we introduce the Segmentation via the Constrained Split Augmented Lagrangian Shrinkage Algorithm (SegSALSA). The effectiveness of the proposed methodology is illustrated with hyperspectral image segmentation.",
keywords = "alternating optimization, Constrained Split Augmented Lagrangian Shrinkage Algorithm (SALSA), hidden fields, hidden Markov measure fields, Image segmentation",
author = "Filipe Condessa and Jos{\'e} Bioucas-Dias and Jelena Kovacevic",
year = "2017",
month = "10",
day = "19",
doi = "10.1109/WHISPERS.2014.8077490",
language = "English (US)",
volume = "2014-June",
booktitle = "2014 6th Workshop on Hyperspectral Image and Signal Processing",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - Supervised hyperspectral image segmentation

T2 - A convex formulation using hidden fields

AU - Condessa, Filipe

AU - Bioucas-Dias, José

AU - Kovacevic, Jelena

PY - 2017/10/19

Y1 - 2017/10/19

N2 - Image segmentation is fundamentally a discrete problem. It consists of finding a partition of the image domain such that the pixels in each element of the partition exhibit some kind of similarity. The optimization is obtained via integer optimization which is NP-hard, apart from few exceptions. We sidestep from the discrete nature of image segmentation by formulating the problem in the Bayesian framework and introducing a hidden set of real-valued random fields determining the probability of a given partition. Armed with this model, the original discrete optimization is converted into a convex program. To infer the hidden fields, we introduce the Segmentation via the Constrained Split Augmented Lagrangian Shrinkage Algorithm (SegSALSA). The effectiveness of the proposed methodology is illustrated with hyperspectral image segmentation.

AB - Image segmentation is fundamentally a discrete problem. It consists of finding a partition of the image domain such that the pixels in each element of the partition exhibit some kind of similarity. The optimization is obtained via integer optimization which is NP-hard, apart from few exceptions. We sidestep from the discrete nature of image segmentation by formulating the problem in the Bayesian framework and introducing a hidden set of real-valued random fields determining the probability of a given partition. Armed with this model, the original discrete optimization is converted into a convex program. To infer the hidden fields, we introduce the Segmentation via the Constrained Split Augmented Lagrangian Shrinkage Algorithm (SegSALSA). The effectiveness of the proposed methodology is illustrated with hyperspectral image segmentation.

KW - alternating optimization

KW - Constrained Split Augmented Lagrangian Shrinkage Algorithm (SALSA)

KW - hidden fields

KW - hidden Markov measure fields

KW - Image segmentation

UR - http://www.scopus.com/inward/record.url?scp=85007469567&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85007469567&partnerID=8YFLogxK

U2 - 10.1109/WHISPERS.2014.8077490

DO - 10.1109/WHISPERS.2014.8077490

M3 - Conference contribution

VL - 2014-June

BT - 2014 6th Workshop on Hyperspectral Image and Signal Processing

PB - IEEE Computer Society

ER -