Markov random field based on Kullback-Leibler divergence and its applications to geo-spatial image segmentation

被引:0
作者
Nishii, R [1 ]
机构
[1] Hiroshima Univ, Fac Integrated Arts & Sci, Higashihiroshima 7398521, Japan
来源
6TH WORLD MULTICONFERENCE ON SYSTEMICS, CYBERNETICS AND INFORMATICS, VOL XVII, PROCEEDINGS: INDUSTRIAL SYSTEMS AND ENGINEERING III | 2002年
关键词
divergence; ICM; Mahalanobis distance; MAP; MRF; multispectral data;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
We consider segmentation of geo-spatial images by Markov random fields. MRFs whose distribution is specified by the Kullback-Leibler divergence between class-conditional densities are introduced for the spatial model. It is shown that the model is a natural extension of Switzer's assumption, and the estimation method of the parameters is established by maximizing the pseudo likelihood. The method is applied to benchmark data, and it shows a good performance.
引用
收藏
页码:399 / 405
页数:7
相关论文
共 11 条
[1]  
BESAG J, 1986, J R STAT SOC B, V48, P259
[2]  
Chiles J.-P., 1999, WILEY SER PROB STAT
[3]  
Cressie N, 1993, STAT SPATIAL DATA
[4]   STOCHASTIC RELAXATION, GIBBS DISTRIBUTIONS, AND THE BAYESIAN RESTORATION OF IMAGES [J].
GEMAN, S ;
GEMAN, D .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1984, 6 (06) :721-741
[5]   Combination of neural and statistical algorithms for supervised classification of remote-sensing images [J].
Giacinto, G ;
Roli, F ;
Bruzzone, L .
PATTERN RECOGNITION LETTERS, 2000, 21 (05) :385-397
[6]   Statistical pattern recognition: A review [J].
Jain, AK ;
Duin, RPW ;
Mao, JC .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2000, 22 (01) :4-37
[7]  
MARDIA KV, 1987, COMMUN STAT THEORY, V16, P3749
[8]   SPATIAL DISCRIMINATION AND CLASSIFICATION MAPS [J].
MARDIA, KV .
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS, 1984, 13 (18) :2181-2197
[9]  
McLachlan, 2004, DISCRIMINANT ANAL ST
[10]   Accuracy and inaccuracy assessments in land-cover classification [J].
Nishii, R ;
Tanaka, S .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 1999, 37 (01) :491-498