基于小波域隐马尔可夫树模型的多光谱遥感影像纹理分割技术研究  

Multi-spectrum Remote Sensing Image Texture Segmentation Research Based on Wavelet-domain Hidden Markov Tree Models

作　　者：彭玲[1] 赵忠明[1] 杨健[1] 马江林[1] 

机构地区：[1]中国科学院遥感应用研究所,北京100101

出　　处：《武汉理工大学学报（交通科学与工程版）》2006年第4期561-564,共4页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：国家863计划项目资助(批准号:2003AA135010)

摘　　要：遥感图像纹理信息丰富,需要进行有效的纹理特征描述.小波域隐马尔可夫树(HM T)模型提供了一种有效的纹理特征提取方法.简述了HM T模型及分割思想,并在此基础上针对多光谱遥感图像,探索更好的纹理描述方法.通过适当的数学变换,得到多光谱遥感影像在不同角度下的描述:H—S—I分量和最大主成分分量.研究比较各分量特性并应用HM T模型,得到不同分量下纹理分割结果.实验表明,利用多光谱信息可有效地提高分割质量.Remote sensing image has rich texture information and need to process effect feature description. Transformation-based model research gives the basic texture feature extraction method. This paper describes the thought of the model establishing and its segmentation, searches the further resolving approach aiming at multi-spectral remote sensing image. Using appropriate mathematical transformation, the description of different aspect ：H-S-I and the first principal components is got. Studying and comparing the characteristic of every component, applying HMT to each of them, the segmented resuits are gained. It seems that the multi-spectral information improves the segmentation effect.

关 键 词：纹理特征 小波域隐马尔可夫树模型(HMT) 多光谱 图像分割 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]