首页 | 本学科首页   官方微博 | 高级检索  
     

面向对象的高分辨率遥感影像城市地物信息提取研究
引用本文:李小强,李金平,甘甜. 面向对象的高分辨率遥感影像城市地物信息提取研究[J]. 平顶山工学院学报, 2014, 0(2): 63-68
作者姓名:李小强  李金平  甘甜
作者单位:云南师范大学旅游与地理科学学院;
摘    要:以沈阳东陵区IKONOS影像为实验数据,采用面向对象分类,构建包含光谱特征、几何特征、纹理特征、拓扑关系等规则的知识库,对城市地物信息进行提取,并将其与基于像素方法的分类结果进行对比分析。研究结果表明:基于多尺度分割的面向对象的分类方法可以有效地避免传统的基于像素分类时出现的"椒盐"现象,分类结果更加符合人类的思维方式,更接近真实值,总体分类精度达到92.5%,比基于像素分类方法更适合作为城市地物专题数据库更新的有效方法。

关 键 词:面向对象  高分辨率遥感  多尺度分割  隶属度函数

Research on high spatial resolution remote sensing image information extraction of urban features based on object-oriented
LI Xiao-qiang,LI Jin-ping,GAN Tian. Research on high spatial resolution remote sensing image information extraction of urban features based on object-oriented[J]. Journal of Pingdingshan Institute of Technology, 2014, 0(2): 63-68
Authors:LI Xiao-qiang  LI Jin-ping  GAN Tian
Affiliation:( School of Tourism and Geography, Yunnan Normal University, Kunming 650500, China)
Abstract:The information extraction technology based on object-oriented is elaborated in this paper. We choose the IKONOS image of Dongling District, Shenyang as experimental data. Based on object-oriented in-formation extraction technology and constructing a knowledge base containing rules of spectral features, geome-try features, texture features, topology, etc. , the method of urban features information extraction is studied. The classification results are compared with the pixel-based classifications. Experimental results show that object-oriented classification based on multi-resolution segmentation can effectively avoid "pepper and salt"phenomenon occurring in pixel-based classification. The classification results are more consistent with the hu-man way of thinking and closer to the real surface features. The overall classification accuracy of object-orien-ted information extraction reaches 92. 5%. Compared with the pixel-based classification method, object-orien-ted information extraction is more suitable for urban features thematic database updating.
Keywords:object-oriented  high spatial resolution remote sensing  multi-resolution segmentation  member-ship function
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号