基于ENVI的遥感图像决策树分类

Decision Tree Classification of Remote Sensing Images Based on ENVI

传统遥感图像分类方法一般是基于概率统计,然而人们一直致力于提高分类精度的研究。本文利用ENVI5.0对研究区分别进行了最大似然法、ISODATA法、决策树三种遥感图像分类;首先对三种分类方法进行简单阐述,然后图像预处理,选取训练样本,最后进行分类。通过数据比较和图表分析,可以看出这三种分类方法中,决策树分类精度最高,最大似然分类次之,ISODATA分类精度最低。决策树分类法可以有效地提高图像分类的精度。

The traditional remote sensing image classification method is based on probability statistics,However,people have always been working on improving the classification accuracy.Using ENVI5.0respectively in the studied area has been carried on with the maximum likelihood method,ISODATA method,three kinds of remote sensing image classification methods.Firstly,three classification methods are simply described,then image preprocessing,selection of training samples,and finally classification.Through data comparison and chart analysis,we can see that the classification accuracy of decision tree is the highest,that of maximum likelihood classification is the second,and the accuracy of ISODATA is the lowest.This also means that,the decision tree classification can effectively improve the accuracy of image classification.