Log in to save to my catalogue

高光谱影像小样本分类的图卷积网络方法

高光谱影像小样本分类的图卷积网络方法

https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_wanfang_journals_chxb202110009

高光谱影像小样本分类的图卷积网络方法

About this item

Full title

高光谱影像小样本分类的图卷积网络方法

Publisher

Beijing: Surveying and Mapping Press

Journal title

Ce hui xue bao, 2021-10, Vol.50 (10), p.1358-1369

Language

Formats

Publication information

Publisher

Beijing: Surveying and Mapping Press

More information

Scope and Contents

Contents

现有的基于卷积神经网络的高光谱影像分类方法通常对影像的规则正方形区域进行卷积,无法普遍适应具有不同地物分布和几何外观的影像局部区域,因此在小样本情况下的分类性能较差,而图卷积网络能对图拓扑信息所代表的不规则影像区域进行卷积。为此,本文提出基于图卷积网络的高光谱影像分类方法。该方法在构建拓扑图的过程中考虑了影像的空间光谱信息,并利用图卷积网络聚合邻居节点的特征信息。在Pavia大学、Indian Pines和Salinas 3个数据集上的试验结果表明,该方法能在训练样本较少的情况下取得较高的分类精度。

Alternative Titles

Full title

高光谱影像小样本分类的图卷积网络方法

Authors, Artists and Contributors

Identifiers

Primary Identifiers

Record Identifier

TN_cdi_wanfang_journals_chxb202110009

Permalink

https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_wanfang_journals_chxb202110009

Other Identifiers

ISSN

1001-1595

E-ISSN

1001-1595

DOI

10.11947/j.AGCS.2021.20200155

How to access this item