针对不同融合算法的遥感影像分类方法研究 - 图文

本 科 毕 业 设 计（论文）

题目 针对不同融合算法的遥感影像

分类方法研究

院（系部）测绘与国土信息工程学院 专业名称 地理信息系统 年级班级 12-02 学生姓名 邹磊 指导教师 刘培

2016年 6月 6日

河南理工大学本科毕业论文

摘要

近年来，随着遥感技术的迅速发展，人们可以获得并使用数量丰富且各具特色的多源遥感影像数据。这些来自同一或不同传感器的影像数据各具自己的优势和不足之处，最为直接的例子就是多光谱影像数据具有良好的光谱特征，这可以帮助我们更好的分清地物的类别，但它的空间分辨率相对较差；而全色影像数据拥有较高空间分辨率，可以使我们很方便的分清地物细节和纹理特征，但其光谱特征很差。这意味着二者在空间分辨率和光谱分辨率两个方面就形成了一定的优势互补性。因此，如何有效地整合这些数据以获得更为丰富精准的信息成为遥感领域重要的研究方向，而遥感融合技术就是因此而迅速发展并逐步完善的一个重要解决方法。遥感影像融合是对同一或不同遥感器的影像进行融合处理的过程，其基本思想是按一定的规则或算法对遥感影像进行运算处理，生成一幅具有新的空间、波谱、时间特征的合成影像，以此获得比任何单一影像更精确、更丰富的信息。本文将对传统像素级遥感影像融合算法PCA、Brovey、Gram—Schmidt变换等进行实验的同时，重点研究处理一种IHS变换和小波变换相结合的新融合算法，并对这些融合结果从定性和定量两方面进行评价，最后利用支持向量机（SVM）分类器以相同的训练区分别对融合前的原始影像及不同融合结果影像进行典型地物分类提取，并使用混淆矩阵方法对分类结果进行整体精度评价。最终实验表明，新算法融合的结果不仅具有比传统融合算法更丰富的信息，更好的光谱特征和较高的空间分辨率，而且在联合熵和分类精度测试方面的表现也很良好。这充分证明了新方法具有很强的优越性和实用性。

关键词：影像融合；传统算法；IHS变换；小波变换；影像分类

1

河南理工大学本科毕业论文

Abstract

In recent years, with the rapid development of remote sensing technology, people can obtain and use the multi-source remote sensing image data which is abundant and distinctive.These image data from the same or different sensors have their own advantages and shortcomings, and the most direct example is the multi spectral image data has good spectral characteristics, which can help us distinguish the type of the feature better. But its spatial resolution is relatively poor; While the panchromatic image data has a high spatial resolution, so that it is very convenient for us to distinguish objects in detail and texture feature.But its spectral characteristics is very poor.This means that the two form certaincomplementary advantages in both aspects of spatial resolution and spectral resolution. Therefore, how to effectively integrate these data to obtain more precise information has become an important research direction in the field of remote sensing, and therefore the remote sensing fusion technology isdeveloping rapidly and becomes a method with gradual improvement.Remote sensing image fusion is the process of image fusion of the same or different remote sensors. The basic idea of remote sensing image is according to certain rules or algorithms for remote sensing image processing,then it will generate a composite image with a new space, spectrum and time characteristics, in order to get more accurate and more abundant information than any single image. These thesis will carry out experiment on the traditional pixel level remote sensing images fusion algorithm of PCA, Brovey, Gram-Schmidt transformation experiment, at the same time, the thesis will focus on the process of IHS transformation and a new fusion algorithm combining wavelet transformation.And then the thesis will evaluate the fusion results from two aspects of qualitative and quantitative. Finally the thesiswill use the support vector machine (SVM) to classify and extract typical objects from the original image before fusion and different image fusion results in the same training area, and do the overall accuracy evaluation on the classification results..The final experimental results show that the results of new fusion algorithmnot only has more abundant information, high spatial resolution and better spectral characteristicsthan the traditional fusion algorithm,but the joint entropy and the accuracy of test performance is also very good. It proves that the new method has strong superiority and practicability.

KeyWords:Image

Fusion; Traditional

2

Algorithm;IHSTransformation;Wavelet

河南理工大学本科毕业论文

Transformation;Image Classification

3

河南理工大学本科毕业论文

目录

1 绪论 ....................................................................................................................................... 1

1.1 引言 ............................................................................................................................ 1 1.2 遥感影像融合的研究历程及现状 ............................................................................ 1 1.3 本文的工作安排 ........................................................................................................ 2 2 遥感影像融合的理论基础 ................................................................................................... 3

2.1 遥感影像特征 ............................................................................................................ 3

2.1.1 空间分辨率 ..................................................................................................... 3 2.1.2 波谱分辨率 ..................................................................................................... 3 2.1.3 时间分辨率 ..................................................................................................... 4 2.2 遥感影像融合的层次 ................................................................................................ 4

2.2.1 像素级融合 ............................................................................................ …… 此处隐藏：5152字，全部文档内容请下载后查看。喜欢就下载吧 ……