CAJ | 학술논문

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 ENVI FX 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。
채용면향대상적영상제취방법，이 ETM ＋영상작위수거원진행수체제취。수선대영상진행분할，획취영상대상，분할시탐구제취수체시영상적최우분할척도；연후대분할생성적영상대상운용 ENVI FX 공구건립규칙집병대수체진행제취。대비면향대상적방법화전통최대사연분류방법적수체제취결과표명：면향대상방법제취수체적총체정도、Kappa계수、생산자정도、용호정도균명현고우전통방법。
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.