中国地理科学(英文版) ›› 2002, Vol. 12 ›› Issue (1): 68-72.

• 论文 • 上一篇    下一篇

THE METHODS OF EXTRACTING WATER INFORMATION FROM SPOT IMAGE

DU Jin-kang, FENG Xue-zhi, WANG Zhou-long, HUANG Yong-sheng, RAMADAN Elnazir   

  1. Department of Urban and Resource Science, Nanjing University, Nanjing 210093 , P. R. China
  • 收稿日期:2001-11-15 出版日期:2002-03-20 发布日期:2011-12-15
  • 作者简介:DU Jin-kang(1964-),male,a native of Muping County,Shandong Province,associate professor. His research interests include bydrological modeling,application of GIS and Remote Sensing to hydrology.

THE METHODS OF EXTRACTING WATER INFORMATION FROM SPOT IMAGE

DU Jin-kang, FENG Xue-zhi, WANG Zhou-long, HUANG Yong-sheng, RAMADAN Elnazir   

  1. Department of Urban and Resource Science, Nanjing University, Nanjing 210093 , P. R. China
  • Received:2001-11-15 Online:2002-03-20 Published:2011-12-15

摘要:

Some techniques and methods for deriving water information from SPOT-4(XI) image were investigated and discussed in this paper. An algorithm of decision tree (DT) classification which includes several classifiers based on the spectral responding characteristics of water bodies and other objects, was developed and put forward to delineate water bodies. Another algorithm of decision tree classification based on both spectral characteristics and auxiliary information of DEM and slope (DTDS) was also designed for water bodies extraction. In addition, supervised classification method of maximum likelyhood classification (MLC), and unsupervised method of interactive self organizing dada analysis technique (ISODATA) were used to extract waterbodies for comparison purpose. An index was designed and used to assess the accuracy of different methods adopted in the research. Results have shown that water extraction accuracy was variable with respect to the various techniques applied. It was low using ISODATA, very high using DT algorithm and much higher using both DTDS and MLC.

关键词: water body, decision tree algorithm, accuracy assessment

Abstract:

Some techniques and methods for deriving water information from SPOT-4(XI) image were investigated and discussed in this paper. An algorithm of decision tree (DT) classification which includes several classifiers based on the spectral responding characteristics of water bodies and other objects, was developed and put forward to delineate water bodies. Another algorithm of decision tree classification based on both spectral characteristics and auxiliary information of DEM and slope (DTDS) was also designed for water bodies extraction. In addition, supervised classification method of maximum likelyhood classification (MLC), and unsupervised method of interactive self organizing dada analysis technique (ISODATA) were used to extract waterbodies for comparison purpose. An index was designed and used to assess the accuracy of different methods adopted in the research. Results have shown that water extraction accuracy was variable with respect to the various techniques applied. It was low using ISODATA, very high using DT algorithm and much higher using both DTDS and MLC.

Key words: water body, decision tree algorithm, accuracy assessment

中图分类号: