面向对象的ETM+影像分割尺度与水体信息提取  

Object Oriented Method of Water Extraction Based on ETM + Images

作　　者：张成才[1] 李艳桦[1] 姚亮亮[1] 

机构地区：[1]郑州大学水利与环境学院,河南郑州450001

出　　处：《人民黄河》2014年第7期54-56,共3页Yellow River

基　　金：河南省基础研究计划项目(132300410031);郑州市科技局科研计划项目(121PPTG360-6);河南省高校科技创新团队支持计划项目(13IRTSTHN030)

摘　　要：采用面向对象的影像提取方法,以ETM+影像作为数据源进行水体提取。首先对影像进行分割,获取影像对象,分割时探究提取水体时影像的最优分割尺度;然后对分割生成的影像对象运用ENVIFX工具建立规则集并对水体进行提取。对比面向对象的方法和传统最大似然分类方法的水体提取结果表明:面向对象方法提取水体的总体精度、Kappa系数、生产者精度、用户精度均明显高于传统方法。This paper used an object-oriented approach to extract water body from the ETM ＋ imagery. Firstly,it got the image object through the image segmentation and explored the optimal segmentation scale for water extraction;then,it used ENVI FX tools to create rule sets and extract wa-ter body. Comparing with the traditional maximum likelihood classification method,the object-oriented method of water extraction performed signifi-cantly better in the overall accuracy,kappa coefficient,producer accuracy and user accuracy. Therefore,the object-oriented method has obvious advantages and good prospects.

关 键 词：ETM + 影像 面向对象 分割尺度 水体提取 

分 类 号：TV213.4[水利工程—水文学及水资源]