文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Statistical image segmentation using Triplet Markov fields
侵权投诉

Statistical image segmentation using Triplet Markov fields

Statistical image segmentation using Triplet Markov fields

Statistical image segmentation using Triplet Markov fields

Wojciech Pieczynski, Dalila Benboudjema, and Pierre Lanchantin1

Département CITI; Institut National des Télécommunications, Evry, France

ABSTRACT

Hidden Markov fields (HMF) are widely used in image processing. In such models, the hidden random field of interestX=(Xs)s∈Sis a Markov field, and the distribution of the observed random field Y=(Ys)s∈S (conditional on X) isgiven by p(yx)=∏p(ysxs). The posterior distribution p(xy) is then a Markov distribution, which affords different

s∈S

Bayesian processing. However, when dealing with the segmentation of images containing numerous classes withdifferent textures, the simple form of the distribution p(yx)above is insufficient and has to be replaced by a Markovfield distribution. This poses problems, because taking p(yx) Markovian implies that the posterior distribution p(xy),whose Markovianity is needed to use Bayesian techniques, may no longer be a Markov distribution, and so differentmodel approximations must be made to remedy this. This drawback disappears when considering directly theMarkovianity of (X,Y); in these recent "Pairwise Markov Fields (PMF) models, both p(yx) and p(xy) are thenMarkovian, the first one allowing us to model textures, and the second one allowing us to use Bayesian restorationwithout model approximations.

In this paper we generalize the PMF to Triplet Markov Fields (TMF) by adding a third random field U=(Us)s∈S andconsidering the Markovianity of (X,U,Y). We show that in TMF X is still estimable from Y by Bayesian methods.The parameter estimation with Iterative Conditional Estimation (ICE) is specified and we give some numerical resultsshowing how the use of TMF can improve the classical HMF based segmentation.

Keywords: Hidden Markov field, Pairwise Markov field, Triplet Markov field, Iterative Conditional Estimation,statistical image segmentation, unsupervised classification.

1. INTRODUCTION

The modeling by hidden Markov fields (HMF) is widely used in various image processing problems. It consists ofconsidering of two stochastic fields X=(Xs)s∈S and Y=(Ys)s∈S, where the unobservable realizations X=x are ofinterest and have to be estimated from the observed Y=y. In this paper we will focuse on the image segmentationproblem and so we will consider that each Xs takes its values in a finite set of classes ={ω1,...,ωk}, and each Ystakes its values in the set of real numbers R. The spelling “hidden Markov” means that the hidden process X has aMarkov distribution. When the distributions p(yx) of Y conditional on X=x are simple enough, the pair (X,Y)keeps the same Markovian form of distribution, and it is the same for the distribution p(xy) of X conditional onY=y. The Markovianity of p(xy) is crucial because it allows one to estimate the unobservable X=x from theobserved Y=y even in the case of very large set S. One possible form of p(yx), which is frequently used inpractice, is p(yx)=∏p(ysxs) [1, 3, 11, 16, 19, 20, 21, 31]. In spite of its simplicity segmentation results obtained

s∈S

are correct in numerous situations, which indicate a good robustness of such HMF. However, when dealing with imagescontaining numerous textures, such a simple form of p(yx) turns out to be in sufficient. In fact, each of textures has to1 E-mails : Wojciech.Pieczynski@int-evry.fr, doc.guandang.net, Dalila.Benboudjema@int-evry.fr,Pierre.Lanchantin@int-evry.fr

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email