Landuse information extraction in Qingdao based on decision tree classification

The paper first describes the method of decision tree classification and characteristics of domestic and foreign research and elaborated on the basic principles of decision tree classifier and decision tree classifier structure the conditions to be met. On this basis, we extract information from Landsat-ETM⁺ image in September 16, 2000. Then we draw the spectral reflectance curve of the samples from the image, and establish the initial decision tree model. After that, we adjust the initial decision tree model to a satisfactory result. At last; we compared the result of decision tree classification with the result of traditional classification. Studies show that the total accuracy of decision tree method increases by 12.96% and Kappa coefficient by 14.63%.