基于高分辨率遥感影像的城市地表信息提取研究  

Urban Thematic Information Extracting Based on High Resolution Satellite Images

作　　者：余晓敏[1,2] 湛飞并[1] 

机构地区：[1]武汉大学资源与环境科学学院,湖北武汉430079 [2]中国科学院深圳先进技术研究院,广东深圳518055

出　　处：《测绘与空间地理信息》2012年第7期21-24,共4页Geomatics & Spatial Information Technology

基　　金：国家863计划项目(2010AA122202);国家重大国际合作项目(2010 DFA92720);深圳基础研究重点项目(JC201005270334 A)资助

摘　　要：针对高分辨率遥感影像在城市地表信息提取中存在的若干问题,发展了一种基于影像对象最优特征组合的城市地表信息提取方法。该方法首先基于面向对象的思想,抽象出城市地表信息所对应的影像对象的各种特征,然后,基于先验知识和样本分析选择最优特征组合,建立有效的影像对象特征集,最后,采用基于知识规则的模糊逻辑分类器快速准确地检测、识别和提取城市各类地表信息。实验结果表明,该方法具有较好的分类精度,其分类结果可为土地利用变化监测和GIS数据库更新提供依据。Aiming at several problems of urban thematic information extracting based on high resolution remote sensing image,a method of urban thematic information extracting based on best feature combination of image objects is developed.The method first Abstracts various features of image objects that urban thematic information correspond to based on object-oriented image analysis.Then the best feature set of image objects is built based on prior knowledge and samples analysis.Finally urban thematic information is identified well and truly by fuzzy logic classifier based on knowledge rules.The experiment results confirm feasibilities of the method.And the classification results can also be used for GIS database updating and land use change detection.

关 键 词：高分辨率影像 面向对象 特征组合 信息提取 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]