Classification of hyperspectral images based on a convolutional neural network and spectral sensitivity

doi:10.1631/jzus.A1900085

CORC > 成都山地灾害与环境研究所 > 中国科学院水利部成都山地灾害与环境研究所 > 山地表生过程与生态调控重点实验室

	Classification of hyperspectral images based on a convolutional neural network and spectral sensitivity
	Ye Cheng-ming 1,5; Liu Xin 5; Xu Hong 1,4; Ren Shi-cong 1,4; Li Yao 3; Li Jonathan 2
刊名	JOURNAL OF ZHEJIANG UNIVERSITY-SCIENCE A
	2020
卷号	21 期号:3 页码:240-248
关键词	Hyperspectral imaging Deep learning Convolutional neural network (CNN) Spectral sensitivity TP751
ISSN号	1673-565X
DOI	10.1631/jzus.A1900085
其他题名	基于卷积神经网络和光谱敏感度的高光谱影像分类方法
通讯作者	Liu, Xin(astluxn@outlook.com)
产权排序	3
文献子类	Article
英文摘要	In recent years, deep learning methods have gradually come to be used in hyperspectral imaging domains. Because of the peculiarity of hyperspectral imaging, a mass of information is contained in the spectral dimensions of hyperspectral images. Also, different objects on a land surface are sensitive to different ranges of wavelength. To achieve higher accuracy in classification, we propose a structure that combines spectral sensitivity with a convolutional neural network by adding spectral weights derived from predicted outcomes before the final classification layer. First, samples are divided into visible light and infrared, with a portion of the samples fed into networks during training. Then, two key parameters, unrecognized rate (delta) and wrongly recognized rate (gamma), are calculated from the predicted outcome of the whole scene. Next, the spectral weight, derived from these two parameters, is calculated. Finally, the spectral weight is added and an improved structure is constructed. The improved structure not only combines the features in spatial and spectral dimensions, but also gives spectral sensitivity a primary status. Compared with inputs from the whole spectrum, the improved structure attains a nearly 2% higher prediction accuracy. When applied to public data sets, compared with the whole spectrum, on the average we achieve approximately 1% higher accuracy.
电子版国际标准刊号	1862-1775
资助项目	Strategic Priority Research Program of the Chinese Academy of Sciences, China[XDA23090203] ; National Key Technologies Research and Development Program of China, China[2016YFB0502600] ; Key Program of Sichuan Bureau of Science and Technology, China[2018SZ0350]
WOS研究方向	Engineering ; Physics
语种	英语
出版者	ZHEJIANG UNIV
WOS记录号	WOS:000520960900007
资助机构	Strategic Priority Research Program of the Chinese Academy of Sciences, China ; National Key Technologies Research and Development Program of China, China ; Key Program of Sichuan Bureau of Science and Technology, China
内容类型	期刊论文
源URL	[http://ir.imde.ac.cn/handle/131551/34131]
专题	成都山地灾害与环境研究所_山地表生过程与生态调控重点实验室
通讯作者	Liu Xin
作者单位	1.Chongqing Engineering Research Center of Automatic Monitoring for Geological Hazards, Chongqing 401120, China; 2.Department of Geography and Environmental Management, University of Waterloo, Waterloo, N2L 3G1, Canada 3.Key Laboratory of Mountain Hazards and Earth Surface Process, Chinese Academy of Sciences, Chengdu 610041, China; 4.National Breeding Base of Technology and Innovation Platform for Automatic-monitoring of Geologic Hazards, Chongqing Institute of Geology and Mineral Resources, Chongqing 401120, China; 5.Key Laboratory of Earth Exploration and Information Technology of Ministry of Education, Chengdu University of Technology, Chengdu 610059, China;
推荐引用方式 GB/T 7714	Ye Cheng-ming,Liu Xin,Xu Hong,et al. Classification of hyperspectral images based on a convolutional neural network and spectral sensitivity[J]. JOURNAL OF ZHEJIANG UNIVERSITY-SCIENCE A,2020,21(3):240-248.
APA	Ye Cheng-ming,Liu Xin,Xu Hong,Ren Shi-cong,Li Yao,&Li Jonathan.(2020).Classification of hyperspectral images based on a convolutional neural network and spectral sensitivity.JOURNAL OF ZHEJIANG UNIVERSITY-SCIENCE A,21(3),240-248.
MLA	Ye Cheng-ming,et al."Classification of hyperspectral images based on a convolutional neural network and spectral sensitivity".JOURNAL OF ZHEJIANG UNIVERSITY-SCIENCE A 21.3(2020):240-248.