基于集成学习的高光谱遥感影像分类综述  

A Review of Hyperspectral Remote Sensing Image Classification Based on Ensemble Learning

作　　者：虞瑶 高涵 陶旸[1] 王圣尧[1] YU Yao;GAO Han;TAO Yang;WANG Shengyao(Provincal Geomatics Centre of Jiangsu,Nanjing 210013,China)

机构地区：[1]江苏省基础地理信息中心,江苏南京210013

出　　处：《测绘与空间地理信息》2023年第4期49-52,60,共5页Geomatics & Spatial Information Technology

基　　金：江苏省自然资源科技项目(2021041)资助。

摘　　要：高光谱遥感影像光谱维度高、数据量大且波段间冗余信息量大,利用集成学习算法可有效地提升高光谱遥感影像的识别精度。本文首先概述了高光谱遥感影像的分类任务和目前存在的问题,其次介绍了集成学习分类算法的原理,系统性阐述了高光谱遥感影像分类中动态集成和静态集成算法的研究现状,并提出了有待进一步研究的问题。Hyperspectral remote sensing images have high spectral dimensions,a large amount of data and a large amount of redundant information between bands.The use of ensemble learning algorithms can improve the recognition accuracy of hyperspectral remote sensing images effectively.This paper firstly summarizes the classification tasks and current problems of hyperspectral remote sensing images.Secondly,it introduces the principles of ensemble learning classification algorithms,systematically explain the research status of dynamic ensemble and static ensemble algorithms in hyperspectral remote sensing image classification,and puts forward questions that need to be further studied.

关 键 词：集成学习 高光谱遥感分类 动态集成 静态集成 

分 类 号：P237[天文地球—摄影测量与遥感]