Hyperspectral Image Classification Based on Gaussian Linear Process and Multi-Neighborhood Optimization

Qin Yang; Xiao hua; Luo Kaiqing

doi:10.3788/LOP57.021013

Laser & Optoelectronics Progress, Volume. 57, Issue 2, 21013(2020)

Hyperspectral Image Classification Based on Gaussian Linear Process and Multi-Neighborhood Optimization

Qin Yang, Xiao hua, and Luo Kaiqing^*

School of Physics and Telecommunication Engineering of China, South China Normal University, Guangzhou, Guangdong 510006, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(6)

Fig. 1. Surface coverage category plane of each database. (a) Indian Pines database; (b) Pavia University database; (c) Salinas database

Download full size

Fig. 2. Classification results of different algorithms on Indian Pines database. (a) Original image; (b) LDA_MLR; (c) GF_LDA_MLR; (d) GF_LDA_MLR_3M

Download full size

Table 1. RGB channel values represented by each category in database
View table
Table 1. RGB channel values represented by each category in database
Color
R 128 0 128 0 128 0 128 64 192 64 192 64 192 64 192 0
G 0 128 128 0 0 128 128 0 0 128 128 0 0 128 128 64
B 0 0 0 128 128 128 128 0 0 0 0 128 128 128 128 0

Table 2. Classification accuracy of each algorithm with MLR as classifier

View table

Table 2. Classification accuracy of each algorithm with MLR as classifier

Database	MLR		LDA_MLR		GF_LDA_MLR		GF_LDA_MLR_3M
Database	OA /%	Kappa	OA /%	Kappa	OA /%	Kappa	OA /%	Kappa
Indian Pines	60.77	0.5371	72.85	0.6896	86.20	0.8422	99.05	0.9891
Pavia University	78.27	0.6965	84.16	0.7881	95.16	0.9357	99.38	0.9918
Salinas	82.82	0.8067	91.45	0.9049	95.63	0.9513	98.88	0.9875

Table 3. Classification results of each spectral-spatial joint method on three databases

View table

Table 3. Classification results of each spectral-spatial joint method on three databases

Database	RBF-SVM		CNN		3DCNN		GF_LDA_MLR_3M
Database	OA /%	Kappa	OA /%	Kappa	OA /%	Kappa	OA /%	Kappa
Indian Pines	77.80	0.7450	85.40	0.8430	92.80	0.9190	94.59	0.9381
Pavia University	90.30	0.8710	93.00	0.9070	95.20	0.9380	98.28	0.9771
Salinas	89.30	0.8800	92.30	0.9140	97.90	0.9770	98.03	0.9781

Table 4. Running time of each method on three databases (training time /test time)s
View table
Table 4. Running time of each method on three databases (training time /test time)s
Database RBF-SVM CNN 3DCNN GF_LDA_MLR_3M
Indian Pines 0.4/1.2 220.7/0.5 2690.2/16.0 1.56/0.56
Pavia University 0.5/3.5 371.8/1.2 1979.0/31.4 1.62/2.30
Salinas 0.4/2.7 165.1/0.7 1157.7/28.1 6.13/2.81

Tools

Get Citation

Copy Citation Text

Qin Yang, Xiao hua, Luo Kaiqing. Hyperspectral Image Classification Based on Gaussian Linear Process and Multi-Neighborhood Optimization[J]. Laser & Optoelectronics Progress, 2020, 57(2): 21013

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Image Processing

Received: Mar. 26, 2019

Accepted: --

Published Online: Jan. 3, 2020

The Author Email: Luo Kaiqing (1573604868@qq.com)

DOI:10.3788/LOP57.021013

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. RGB channel values represented by each category in database

Table 1. RGB channel values represented by each category in database

Table 2. Classification accuracy of each algorithm with MLR as classifier

Table 2. Classification accuracy of each algorithm with MLR as classifier

Table 3. Classification results of each spectral-spatial joint method on three databases

Table 3. Classification results of each spectral-spatial joint method on three databases

Table 4. Running time of each method on three databases (training time /test time)s

Table 4. Running time of each method on three databases (training time /test time)s