Research on extraction of vegetation information from remote sensing images based on knowledge rules

The paper is based on the research of the northern area of Haidian District in Beijing, and brings forth a method to build knowledge rules based on remote sensing images spectral features and DEM elevation features. The vegetation Information in this research area is extracted and classified And the courses and results of this method and the supervised classification method are compared during this research. The result shows that the classification results of this decision tree method are superior to the supervised classification method; this method can eliminate the deficiency of classification by using only spectral information and it can effectively solve the ground object mix-separate phenomenon to achieve high classification accuracy.