摘要遥感影像的分辨率得到大幅度的提高,影像中不仅地物的光谱特征更丰富， 而且更加突出了地物的内部结构、纹理、边缘等信息，高分辨率的遥感卫星数据 在遥感的应用领域中将发挥出越来越大的作用。69978

本文主要针对高分辨率遥感影像的特点对高分辨率影像分类方法进行研究。 在遥感影像分类中目前应用较多的方法有最大似然法、决策树法、神经网络法等 分类算法，其中决策树法因其计算方便，应用灵活、直观，并且能较为准确地表 达高分辨率遥感影像信息，越来越受到更多研究人员的重视。本论文对决策树分 类算法进行研究，并利用决策树分类中的 CART 算法实现了 QuickBird 影像的分 类，并将分类结果与遥感影像分类中常用的最大似然法进行了比较分析与评价。

本论文有图 7 幅，表 6 个，参考文献 19 篇。

毕业论文关键词：决策树 高分辨率遥感影像 遥感分类

Research on High Resolution Remote Sensing Image Classification based on Decision Tree

Abstract Resolution remote sensing image has been greatly improved, the spectral characteristics of the image feature not only richer, but also more prominent internal structure, texture, edges and other surface features of information, high-resolution remote sensing satellite data in the field of remote sensing applications It will play an increasingly large role.

In this paper, the characteristics of high resolution satellite images to study high-resolution image classification method. Remote sensing image classification methods are currently used more maximum likelihood method, decision tree, neural network classification algorithm method, wherein the decision tree method because of its ease of calculation, flexible, intuitive, and can accurately express resolution rate of remote sensing image information, more and more attention of researchers. The thesis of the decision tree classification algorithm research, and the CART decision tree classification algorithm classifies QuickBird imagery and classification results with remote sensing image classification commonly used in the maximum likelihood method were compared and evaluated.

In this paper, there are pictures 7 pieces, a table 6,19 chapter references.

Key  Words:  Decision Tree High  Resolution  Remote  Sensing Image Remote Sensing Classification

目 录

摘 要 I

Abstract II

目 录 III

图清单 V

表清单 V

1 绪论 1

1.1 课题研究背景和研究意义 1

1.2 高分辨率遥感影像分类研究现状 1

1.3 决策树分类算法研究 2

1.4 决策树在高分辨率遥感影像中的研究进展 3

1.5 本文组织结构 4

2 研究区概况及研究资料 5

2.1 研究区位置及资料 5

2.2 研究方法及流程图 6

3 基于决策树技术的高分辨率遥感图像数据处理 8

3.1 数据简介 8

3.2 数据预处理