Parameter estimation in Markov random field image modeling with imperfect observations.: A comparative study

被引：31

作者：

Ibáñez, MV ^{[1
]}

Simó, A ^{[1
]}

机构：

[1] Univ Jaume 1, Dept Math, Castellon de La Plana 12071, Spain

来源：

PATTERN RECOGNITION LETTERS | 2003年 / 24卷 / 14期

关键词：

parameter estimation; MRF models; unsupervised image analysis; Monte Carlo likelihood; simulation study; MCMC methods;

D O I：

10.1016/S0167-8655(03)00067-9

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Many algorithms in unsupervised image analysis are based on Markov random fields, and parameter estimation plays an important role. Two difficulties are usually present: the presence of unobserved data and the fact that the normalizing constant of the model is unknown. In this paper we show the application to this context of a parameter estimation method which is popular in the point process context. We shortly review other related methods and finally we do a simulation study in order to compare them. (C) 2003 Elsevier B.V. All rights reserved.

引用

页码：2377 / 2389

页数：13

共 50 条

[1]

[Anonymous], 1995, RANDOM FIELDS NETWOR

[2]

[Anonymous], 1985, Computational Statistics Quarterly, DOI DOI 10.1155/2010/874592

[3] GLOBAL AND LOCAL PRIORS, AND THE LOCATION OF LESIONS USING GAMMA-CAMERA IMAGERY [J].

AYKROYD, RG ;

GREEN, PJ .

PHILOSOPHICAL TRANSACTIONS OF THE ROYAL SOCIETY OF LONDON SERIES A-MATHEMATICAL PHYSICAL AND ENGINEERING SCIENCES, 1991, 337 (1647) :323-342

[4] Unsupervised image segmentation using Markov random field models [J].

Barker, SA ;

Rayner, PJW .

PATTERN RECOGNITION, 2000, 33 (04) :587-602

[5] STATISTICAL-ANALYSIS OF NON-LATTICE DATA [J].

BESAG, J .

JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES D-THE STATISTICIAN, 1975, 24 (03) :179-195

[6]

BESAG J, 1986, J R STAT SOC B, V48, P259

[7] Pattern recognition using Markov random field models [J].

Cai, JH ;

Liu, ZQ .

PATTERN RECOGNITION, 2002, 35 (03) :725-733

[8]

CHALMOND B, 1989, PATTERN RECOGN, V15, P115

[9] Unsupervised deconvolution of sparse spike trains using stochastic approximation [J].

Champagnat, F ;

Goussard, Y ;

Idier, J .

IEEE TRANSACTIONS ON SIGNAL PROCESSING, 1996, 44 (12) :2988-2998

[10]

CHELLAPPA R, 1991, MARKOV RANDOM FIELDS

← 1 2 3 4 5 →