遥感影像像斑空间关系分析的非规则无参数马尔可夫随机场模型

针对遥感影像像斑特点和单像斑分析的不足,提出一种非规则马尔可夫随机场模型,将空间关系特征应用于遥感影像像斑的解译。该方法将基于邻接关系定义的随机场与BP神经元网络相结合,通过建立无参数邻域势函数,采用基于类别相邻规律的迭代统计方法,避免了传统吉布斯参数的复杂估计过程,实现了像斑单点势能和邻域势能计算。试验证明,该方法能有效实现像斑自身特征和邻域特征共同参与像斑属性判断,显著提高遥感影像像斑分析效果。...

Full description

Saved in:
Bibliographic Details
Published in测绘学报 Vol. 42; no. 1; pp. 101 - 107
Main Author 龚 舒宁 王琰 陶醉 尹淑玲
Format Journal Article
LanguageChinese
Published 武汉大学遥感信息工程学院,湖北武汉,430079%中国科学院遥感应用研究所,北京,100101%武昌理工学院信息工程学院,湖北武汉,430223 2013
Subjects
Online AccessGet full text
ISSN1001-1595

Cover

More Information
Summary:针对遥感影像像斑特点和单像斑分析的不足,提出一种非规则马尔可夫随机场模型,将空间关系特征应用于遥感影像像斑的解译。该方法将基于邻接关系定义的随机场与BP神经元网络相结合,通过建立无参数邻域势函数,采用基于类别相邻规律的迭代统计方法,避免了传统吉布斯参数的复杂估计过程,实现了像斑单点势能和邻域势能计算。试验证明,该方法能有效实现像斑自身特征和邻域特征共同参与像斑属性判断,显著提高遥感影像像斑分析效果。
Bibliography:segmants; Markov random field; BP neural net; spatial relations; object-oriented; Gibbs parameters
11-2089/P
GONG Yan,SHU Ning,WANG Yan,TAO Zui,YIN Shuling( 1.School of Remote Sensing and Information Engineering,Wuhan University,Wuhan 430079,China;2.Institute of Remote Sensing Applications,Chinese Academy of Sciences,Beijing 100101,China;3.Information Engineering School,Wuchang University of Technology,Wuhan 430223,China )
Spatial relationship feature was applied in segments' interpretation for remote sensing imagery,and a new irregular MRF model was offered aimed at segments' characteristic.This new method combined the random field defined by adjacent relationship with BP neural net,founded a non-parameter energy function assisted by the iterative statistic of neighborhood class label cooccurrence,and it avoided the traditional complicated estimation of Gibbs parameters.Algorithm comparisons were done with multispectral images and high resolution images.Experiments demonstrate the proposed method can co-analyze
ISSN:1001-1595