The Study of Automatically Extracting Water Information in City Zone Based On SPOT5 Image

This article looks at the main city zones as the research area, and studies the method of extracting water information from SPOT5 images. We choose the SPOT5 as the research data as we can get the water and shadow information from the image by setting the dividing line in the SWIR band. By using the spectrum characteristic, the space characteristic and time characteristic, such as index of shape, the decision tree model of automatically extracting water information in a city zone from SPOT5 image to get the water information from the area can be set up. For estimating the precision of the model, the model and supervised classification method in the whole area and in some special zone that has much building shadow are compared. The result tells us that in the whole area there is some improvement between the model and supervised classification method, about 2.5%, especially in the special zone there is large improvement, reaching 11.6%.