Decision Tree classification of remote sensing images based on index
In order to study the advantage of decision tree based on index in land use information extracted, taking Harbin as a resa?mple,there are 20 spectral indexes was calculated,including water index,vegetation,construction and soil type index and so on,using Land?sat-8 multispectral images. According to the validation sample verifies the accuracy of the result of the classification,and then compari?son and analysis the influence of different index on classification accuracy.Based on this,the higher precision index was classified into 5 groups to extract the area land use type,the classification result compared with decision tree classification based on spectral and maxi?mum likelihood method.The result indicates that five groups based on the index of decision tree classification precision were better than the decision tree classification based on spectral and the maximum likelihood classification,one group was highest among five groups, which total precision improved by 2.59% and 9.55%,the Kappa coefficient increased 0.08 and 0.15,respectively.This paper presents the advantage of the decision tree based on index in land use information extraction,which provide basic data for better coordination of Har?bin city of land use and urban expansion.
Decision tree classificationIndexesInformation extractionLand use
NETL
NSTL
万方数据
维普
