Spectrum selection and deep feature fusion based hyperspectral image natural scene classification network

Weilong Guo; Zifei Zhao; Longxuan Kou; Junjie Lu; Shaopan Xiong; Zhuang Zhou; Shengyang Li; Wei Wu

doi:10.1117/12.2588977

18 March 2021 Spectrum selection and deep feature fusion based hyperspectral image natural scene classification network

Weilong Guo, Zifei Zhao, Longxuan Kou, Junjie Lu, Shaopan Xiong, Zhuang Zhou, Shengyang Li, Wei Wu

Proceedings Volume 11780, Global Intelligent Industry Conference 2020; 117800A (2021) https://doi.org/10.1117/12.2588977
Event: Global Intelligent Industry Conference 2020, 2020, Guangzhou, China

Abstract

Hyperspectral image classification plays an important role in many remote sensing applications. However, the high-dimensional characteristics of hyperspectral images and the appropriate feature representations leave it with great challenges. In this article, these difficulties are addressed by developing a Spectrum Selection and Deep Feature Fusion based method. The proposed method has the following contributions: 1) reducing redundant information caused by high-dimension through spectrum selection which is just needed in training phase. 2) extracting the joint spectral-spatial features by deep feature fusion, which effectively improves the accuracy of scene classification. 3) increasing the network attention to scene classes of small number by the Class-Balanced loss function and overcome the influence of unbalanced distribution of experimental data. Experiments results in the Tiangong-1 natural scene images dataset (TG1-NSCD) demonstrate that the effectiveness of our algorithm and the OA is 17% higher than the baseline.

Citation Download Citation

Weilong Guo, Zifei Zhao, Longxuan Kou, Junjie Lu, Shaopan Xiong, Zhuang Zhou, Shengyang Li, and Wei Wu "Spectrum selection and deep feature fusion based hyperspectral image natural scene classification network", Proc. SPIE 11780, Global Intelligent Industry Conference 2020, 117800A (18 March 2021); https://doi.org/10.1117/12.2588977

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available