EMAPs-SMLR融合方法在高光谱图像分类中的应用

46 浏览量更新于2024-08-27 2 收藏 505KB PDF 举报

"本文提出了一种新的高光谱图像(HSI)分类方法，结合了EMAPs(扩展形态学多属性谱)的空间信息提取和SMLR(稀疏多项式逻辑回归)的分类技术，实现了空间光谱特征的有效融合，提高了HSI分类的性能。在实验中，该方法表现出了优越的分类结果。" 高光谱图像分类是遥感领域的关键任务之一，用于识别地物类型。传统的方法通常仅依赖于光谱信息，而忽略了图像的空间信息。针对这一问题，本文提出了一种基于EMAPs和SMLR的新策略，旨在充分利用高光谱图像的空间和光谱特性。 EMAPs(Extended Morphological Multi-Attribute Profiles)算法是本文的核心部分，它是一种有效的空间信息提取工具。EMAPs通过使用多个属性结构替换简单的结构元素，并将它们级联来获取多个结构的属性特征。这种方法可以更全面地捕捉HSI中的复杂空间模式，从而提高后续分类的准确性。通过将EMAPs提取的空间信息与原始光谱信息相结合，形成了一种空间光谱特征融合模型，这种融合有助于增强特征的判别能力。接下来，文章引入了SMLR(Sparse Multinomial Logistic Regression)作为分类器。SMLR是一种适用于处理高维数据和大规模数据集的统计学习方法，尤其适合HSI的特征维度高、样本量大的情况。SMLR基于多项式逻辑回归，但通过学习稀疏模型，能够减少冗余特征，提高分类效率。在HSI分类中，采用基于MLR的多分类器，利用快速算法学习稀疏的模型，以降低计算复杂度并保持高分类精度。实验结果显示，与现有的HSI分类方法相比，提出的EMAPs-SMLR方法在分类性能上表现出显著的优势。这表明，结合空间光谱特征融合和SMLR的策略在高光谱图像分析中具有极大的潜力，为HSI分类提供了一条新的有效途径。未来的研究可能会进一步优化EMAPs和SMLR的参数设置，以及探索它们与其他特征选择或特征提取方法的结合，以提升HSI分类的整体性能。

Hyperspectral Image Classification Based on EMAPs Spatial-Spectral

Features Fusion and SMLR

Rong Ren

a,b

, Wenxing Bao*

School of Computer Science and Engineering, North Minzu University, YinChuan NingXia 750021;

School of computers and information, Hefei University of Technology，HeFei, AnHui 230009

ABSTRACT

This paper presents a novel classification method of hyperspectral image(HSI) based on EMAPs and SMLR. Firstly, we

adopt EMAPs(Extended Morphological multi-Attribute Profiles) algorithm to extract the spatial information of HSI

efficiently, and combine the spectral information to form spatial-spectral features fusion model. EMAPs can replace

simple structural elements with multiple attributes structure and cascade them to obtain attributes feature of multiple

structures. Then we utilize SMLR(Sparse Multinomial Logistic Regression) for HSI classification. SMLR is applicable

to high-dimensional and large data sets. A multiclass classifier based on MLR is adopted, and a fast algorithm is used to

learn a sparse multiclass classifier. Compared with other methods in HSI experiments, our method provides an excellent

result.

Keywords: hyperspectral image, classification, EMAPs, SMLR

1. INTRODUCTION

The advantage of hyperspectral image(HSI) is the combination of imagery and spectrum. The spectral information is

continuous, complete and informative. Hyperspectral remote sensing technology has been widely used in precision

agriculture, forestry monitoring, marine management, and environmental protection etc, and many valuable

achievements has been achieved

Classification is an important content and research focus of HSI analysis. Many classification techniques have been

developed for HSI, includes supervised classification method, such as SVM (Support Vector Machine), maximum

likelihood method, decision tree, minimum distance, artificial neural network classification, and unsupervised

classification method, such as K-Means and ISODATA. Machine learning technique (e.g., SVM or MLR(Multinomial

Logistic Regression)) has shown remarkable success

1,2

for HSI classification because of its superiorities in dealing with

high dimensional data and Hughes phenomenon

3,4

. Recently, sparse MLR(SMLR)

is investigated in many literatures,

and it was applied to HSI classification successfully

6,7

for its advantages of the fast and the applicability to large scale

data. Compare with the frequently-used method, such as SVM, SMLR has achieved better classification effect. However,

if SMLR method only make use of spectral feature, it will lead to outliers or inefficient classification.

In order to solve the problem, spatial-spectral combination classification

8,9,10

technology has become research hot spot

now. Spatial information can describe morphology, texture and edge features of HSI, so it is very significant to improve

HSI classification effect. EMAPs(Extended Morphological multi-Attribute Profiles) is an efficient algorithm to extract

spatial features of HSI.

The previous researches show that EMAPs is a powerful method to extract image features, and

feature extraction of spatial information can improve HSI class separability greatly

11,12

In this paper, we proposed a classification method innovatively based on EMAPs spatial-spectral fusion and SMLR. The

HSI classification experiments showed that the result of our proposed method is better than other methods.

The remainder of this paper is organized as follows. Section II describes the flow of our method and the relevant

algorithms such as EMAPs, SMLR. Section III presents experimental results using the HSI of Pavia University and

Indian Pines. Section IV concludes this paper and puts forward the future research work.

2. RESEARCH METHOD

2.1 The Flow of Proposed Method

Firstly, EMAPs algorithm is used to extract the spatial features of HSI. EMAPs can reconstruct the spatial and spectral

information of HSI from different directions. Then multi-features fusion model is built through combination of spatial

and spectral features. After random sampling, first-order neighborhood matrix is constructed. RBF ( Radial Basis Function)

下载后可阅读完整内容，剩余7页未读，立即下载

weixin_38718415

粉丝: 11
资源: 951

EMAPs-SMLR融合方法在高光谱图像分类中的应用

图像直方图的规定化处理（SML算法）

MLR和稀疏MLR分类器的并行算法：该算法可用于多项式和稀疏多项式逻辑回归分类器的参数估计。-matlab开发

基于形态学属性剖面的高光谱图像分类特征提取

电气工程及其自动化 (2).docx

vs-Community2017

数据科学与大数据技术 (10).docx

高跟鞋检测24-YOLOv8数据集合集.rar

JSP基于SSM新闻发布系统网站设计毕业源码案例设计.zip

配电网自动化技术—配电网馈线监控终端.pptx

学生毕业离校系统-管理系统-毕业设计源码.zip

最新资源