基于极限学习机的高光谱图像监督分类
摘要高光谱图像监督分类在土地分类、水质监测等领域具有重要应用价值，其关键是少量训练样本下的分类监督性能。本文围绕高光谱图像监督分类的极限学习机方法进行研究。论文的主要工作为：
(1)研究了支持向量机监督分类方法，并以此作为性能测试的标准。49133
(2)研究了极限学习机的高光谱图像分类监督分类原理及算法实现，并对其改进。结合空谱信息，本文利用RBF核，设计了一种RBF-KELM监督分类算法。
(3)综合实现了SVM，ELM和RBF-KELM等算法，应用于高光谱图像监督分类，综合进行了分类实验和性能评测。
实验结果表明：结合空谱信息的RBF-KELM算法有较好的分类精度和不错的分类速度。
关键词 高光谱图像 分类 特征空间 空谱信息的联合 极限学习机 支持向量机
毕业论文设计说明书外文摘要
Title Supervised Classification of Hyperspectral Image Based on Extreme Learning Machine
Abstract
Supervised Hyperspectral image classification has important applications in land classification and in water quality monitoring,where the key lies in how to boost the classification performance with limited number of training samples. This paper focuses on the Extreme Learning Machine method in supervised hyperspectral image classification and the main work includes:
(1) Investigated the supervised SVM classification method, based on which to establish the performance standard.
(2) Investigated the principles how Extreme Learning Machine can be to applied to hyperspectral image classification and implemented the improve algorithm. In combined with the space spectral information, I designed an supervised RBF-KELM classification algorithm utilizing the RBF kernel.
(3) Implemented the SVM, ELM as well as RBF-KELM algorithms respectively and applied them to supervised hyperspectral image classification, followed by the classification experiments and the performance evaluation.
The experiments suggested that: the RBF-KELM algorithm that integrates the space spectrum information has superior classification precision as well as speed.
Keywords Hyperspectral image, classification, Feature space, Space Spectrum integration information, Extreme Learning Machine, Support Vector Machine
目 次
1 绪论 1
1.1 研究背景和意义 1
1.3 本文的主要内容 4
2 支持向量机的分类原理 6
2.1 引言 6
2.2 支持向量机的基本原理 6
2.3 基于MH预测的SVM高光谱图像分类模型 12
2.4 本章小结 14
3 基于核极限学习的高光谱图像监督分类 15
3.1 引言 15
3.2 极限学习机的相关原理 15
3.3 基于空间相关性的MH预测KELM的高光谱图像分类 21
3.4 本章小结 25
4 实验与结果分析 26
4.1 定性与定量评测的标准说明