Segmentation of color images via reversible jump MCMC sampling

Research output: Contribution to journalArticle

38 Citations (Scopus)

Abstract

Reversible jump Markov chain Monte Carlo (RJMCMC) is a recent method which makes it possible to construct reversible Markov chain samplers that jump between parameter subspaces of different dimensionality. In this paper, we propose a new RJMCMC sampler for multivariate Gaussian mixture identification and we apply it to color image segmentation. For this purpose, we consider a first order Markov random field (MRF) model where the singleton energies derive from a multivariate Gaussian distribution and second order potentials favor similar classes in neighboring pixels. The proposed algorithm finds the most likely number of classes, their associated model parameters and generates a segmentation of the image by classifying the pixels into these classes. The estimation is done according to the Maximum A Posteriori (MAP) criterion. The algorithm has been validated on a database of real images with human segmented ground truth.

Original languageEnglish
Pages (from-to)361-371
Number of pages11
JournalImage and Vision Computing
Volume26
Issue number3
DOIs
Publication statusPublished - Mar 3 2008

Fingerprint

Markov processes
Sampling
Color
Pixels
Gaussian distribution
Image segmentation

Keywords

  • Color
  • Markov random fields
  • Normal mixture identification
  • Parameter estimation
  • Reversible jump Markov chain Monte Carlo
  • Simulated annealing
  • Unsupervised image segmentation

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Segmentation of color images via reversible jump MCMC sampling. / Kato, Z.

In: Image and Vision Computing, Vol. 26, No. 3, 03.03.2008, p. 361-371.

Research output: Contribution to journalArticle

@article{2782691560804afa8016935caeada726,
title = "Segmentation of color images via reversible jump MCMC sampling",
abstract = "Reversible jump Markov chain Monte Carlo (RJMCMC) is a recent method which makes it possible to construct reversible Markov chain samplers that jump between parameter subspaces of different dimensionality. In this paper, we propose a new RJMCMC sampler for multivariate Gaussian mixture identification and we apply it to color image segmentation. For this purpose, we consider a first order Markov random field (MRF) model where the singleton energies derive from a multivariate Gaussian distribution and second order potentials favor similar classes in neighboring pixels. The proposed algorithm finds the most likely number of classes, their associated model parameters and generates a segmentation of the image by classifying the pixels into these classes. The estimation is done according to the Maximum A Posteriori (MAP) criterion. The algorithm has been validated on a database of real images with human segmented ground truth.",
keywords = "Color, Markov random fields, Normal mixture identification, Parameter estimation, Reversible jump Markov chain Monte Carlo, Simulated annealing, Unsupervised image segmentation",
author = "Z. Kato",
year = "2008",
month = "3",
day = "3",
doi = "10.1016/j.imavis.2006.12.004",
language = "English",
volume = "26",
pages = "361--371",
journal = "Image and Vision Computing",
issn = "0262-8856",
publisher = "Elsevier Limited",
number = "3",

}

TY - JOUR

T1 - Segmentation of color images via reversible jump MCMC sampling

AU - Kato, Z.

PY - 2008/3/3

Y1 - 2008/3/3

N2 - Reversible jump Markov chain Monte Carlo (RJMCMC) is a recent method which makes it possible to construct reversible Markov chain samplers that jump between parameter subspaces of different dimensionality. In this paper, we propose a new RJMCMC sampler for multivariate Gaussian mixture identification and we apply it to color image segmentation. For this purpose, we consider a first order Markov random field (MRF) model where the singleton energies derive from a multivariate Gaussian distribution and second order potentials favor similar classes in neighboring pixels. The proposed algorithm finds the most likely number of classes, their associated model parameters and generates a segmentation of the image by classifying the pixels into these classes. The estimation is done according to the Maximum A Posteriori (MAP) criterion. The algorithm has been validated on a database of real images with human segmented ground truth.

AB - Reversible jump Markov chain Monte Carlo (RJMCMC) is a recent method which makes it possible to construct reversible Markov chain samplers that jump between parameter subspaces of different dimensionality. In this paper, we propose a new RJMCMC sampler for multivariate Gaussian mixture identification and we apply it to color image segmentation. For this purpose, we consider a first order Markov random field (MRF) model where the singleton energies derive from a multivariate Gaussian distribution and second order potentials favor similar classes in neighboring pixels. The proposed algorithm finds the most likely number of classes, their associated model parameters and generates a segmentation of the image by classifying the pixels into these classes. The estimation is done according to the Maximum A Posteriori (MAP) criterion. The algorithm has been validated on a database of real images with human segmented ground truth.

KW - Color

KW - Markov random fields

KW - Normal mixture identification

KW - Parameter estimation

KW - Reversible jump Markov chain Monte Carlo

KW - Simulated annealing

KW - Unsupervised image segmentation

UR - http://www.scopus.com/inward/record.url?scp=36549047468&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=36549047468&partnerID=8YFLogxK

U2 - 10.1016/j.imavis.2006.12.004

DO - 10.1016/j.imavis.2006.12.004

M3 - Article

AN - SCOPUS:36549047468

VL - 26

SP - 361

EP - 371

JO - Image and Vision Computing

JF - Image and Vision Computing

SN - 0262-8856

IS - 3

ER -