| Nowadays, it is an important way of survey and macro-supervision of water source that water body information can be quickly and accurately extracted from satellite remote sensing images. As for extracting water body information in high mountain and canyon area, there exist plenty of shadow to influence extraction of water body information. It is difficult to identify the difference of spectrum characteristics between water body information and mass terrain shadow. How to decrease or eliminate the influence of mountain body shadow is the key factor to implement automatic extraction of water body information in the research field through computer, and this is also the main research content in this paper.Based on deep analysis about diversified extraction method of remote sensing data, multi different wave band spectral relationship and threshold method, modified normalized difference water index and threshold method, similar vegetation index and threshold method, and decision tree method are all used to identify water body information from ETM in typical high mountain and canyon area-Li Jiang county. At last, evaluation and precision analysis of results using these four ways to automatically extracting water body information are studied, comparing with visual judgment and image vectorization of human-computer interacting interpretation. Mostly research achievements list as following:(1) If threshold method is used by itself, it is very difficult to distinguish the difference between water body and mountain shadow of similar gray value. But multi different wave band spectral relationship and threshold method, modified normalized difference water index and threshold method, similar vegetation index and threshold method, and decision tree method are all effective to eliminate the influence of shadow.(2) In Li Jiang county, multi spectral relationship and threshold method is most effective method to extract water body information and eliminate influence of shadow.
|
PDF Full Text Request