Markov random field texture models for classification

Roman Antosik, David R Scott, Gerald M. Flachs

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

Abstract

Two novel approaches to texture classification based upon stochastic modeling using Markov Random Fields are presented and contrasted. The first approach uses a clique-based probabilistic neighborhood structure and Gibbs distribution to derive the quasi-likelihood estimates of the model coefficients. The second approach uses a least squares prediction error model and error signature analysis to model and classify textures. A new statistic and complexity measure are introduced called the K-nearest neighbor statistic (KNS) and complexity (KNC) which measure the overlap in K-nearest neighbor conditional distributions. Parameter vectors for each model, neighborhood size and structure, performance of the maximum likelihood and K-nearest neighbor decision strategies are presented and interesting results discussed. Results from classifying real video pictures of six cloth textures are presented and analyzed.

Original languageEnglish (US)
Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
EditorsPaul Janota
PublisherPubl by Int Soc for Optical Engineering
Pages48-57
Number of pages10
Volume1301
StatePublished - 1990
Externally publishedYes
EventDigital Image Processing and Visual Communications Technologies in the Earth and Atmospheric Sciences - Orlando, FL, USA
Duration: Apr 18 1990Apr 19 1990

Other

OtherDigital Image Processing and Visual Communications Technologies in the Earth and Atmospheric Sciences
CityOrlando, FL, USA
Period4/18/904/19/90

Fingerprint

textures
Textures
signature analysis
Statistics
statistics
classifying
Maximum likelihood
coefficients
estimates
predictions

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Condensed Matter Physics

Cite this

Antosik, R., Scott, D. R., & Flachs, G. M. (1990). Markov random field texture models for classification. In P. Janota (Ed.), Proceedings of SPIE - The International Society for Optical Engineering (Vol. 1301, pp. 48-57). Publ by Int Soc for Optical Engineering.

Markov random field texture models for classification. / Antosik, Roman; Scott, David R; Flachs, Gerald M.

Proceedings of SPIE - The International Society for Optical Engineering. ed. / Paul Janota. Vol. 1301 Publ by Int Soc for Optical Engineering, 1990. p. 48-57.

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

Antosik, R, Scott, DR & Flachs, GM 1990, Markov random field texture models for classification. in P Janota (ed.), Proceedings of SPIE - The International Society for Optical Engineering. vol. 1301, Publ by Int Soc for Optical Engineering, pp. 48-57, Digital Image Processing and Visual Communications Technologies in the Earth and Atmospheric Sciences, Orlando, FL, USA, 4/18/90.
Antosik R, Scott DR, Flachs GM. Markov random field texture models for classification. In Janota P, editor, Proceedings of SPIE - The International Society for Optical Engineering. Vol. 1301. Publ by Int Soc for Optical Engineering. 1990. p. 48-57
Antosik, Roman ; Scott, David R ; Flachs, Gerald M. / Markov random field texture models for classification. Proceedings of SPIE - The International Society for Optical Engineering. editor / Paul Janota. Vol. 1301 Publ by Int Soc for Optical Engineering, 1990. pp. 48-57
@inproceedings{b806ae7ac6734c89a9613e080cff5778,
title = "Markov random field texture models for classification",
abstract = "Two novel approaches to texture classification based upon stochastic modeling using Markov Random Fields are presented and contrasted. The first approach uses a clique-based probabilistic neighborhood structure and Gibbs distribution to derive the quasi-likelihood estimates of the model coefficients. The second approach uses a least squares prediction error model and error signature analysis to model and classify textures. A new statistic and complexity measure are introduced called the K-nearest neighbor statistic (KNS) and complexity (KNC) which measure the overlap in K-nearest neighbor conditional distributions. Parameter vectors for each model, neighborhood size and structure, performance of the maximum likelihood and K-nearest neighbor decision strategies are presented and interesting results discussed. Results from classifying real video pictures of six cloth textures are presented and analyzed.",
author = "Roman Antosik and Scott, {David R} and Flachs, {Gerald M.}",
year = "1990",
language = "English (US)",
volume = "1301",
pages = "48--57",
editor = "Paul Janota",
booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",
publisher = "Publ by Int Soc for Optical Engineering",

}

TY - GEN

T1 - Markov random field texture models for classification

AU - Antosik, Roman

AU - Scott, David R

AU - Flachs, Gerald M.

PY - 1990

Y1 - 1990

N2 - Two novel approaches to texture classification based upon stochastic modeling using Markov Random Fields are presented and contrasted. The first approach uses a clique-based probabilistic neighborhood structure and Gibbs distribution to derive the quasi-likelihood estimates of the model coefficients. The second approach uses a least squares prediction error model and error signature analysis to model and classify textures. A new statistic and complexity measure are introduced called the K-nearest neighbor statistic (KNS) and complexity (KNC) which measure the overlap in K-nearest neighbor conditional distributions. Parameter vectors for each model, neighborhood size and structure, performance of the maximum likelihood and K-nearest neighbor decision strategies are presented and interesting results discussed. Results from classifying real video pictures of six cloth textures are presented and analyzed.

AB - Two novel approaches to texture classification based upon stochastic modeling using Markov Random Fields are presented and contrasted. The first approach uses a clique-based probabilistic neighborhood structure and Gibbs distribution to derive the quasi-likelihood estimates of the model coefficients. The second approach uses a least squares prediction error model and error signature analysis to model and classify textures. A new statistic and complexity measure are introduced called the K-nearest neighbor statistic (KNS) and complexity (KNC) which measure the overlap in K-nearest neighbor conditional distributions. Parameter vectors for each model, neighborhood size and structure, performance of the maximum likelihood and K-nearest neighbor decision strategies are presented and interesting results discussed. Results from classifying real video pictures of six cloth textures are presented and analyzed.

UR - http://www.scopus.com/inward/record.url?scp=0025535293&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0025535293&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0025535293

VL - 1301

SP - 48

EP - 57

BT - Proceedings of SPIE - The International Society for Optical Engineering

A2 - Janota, Paul

PB - Publ by Int Soc for Optical Engineering

ER -