A Hyperspectral Image Classification Method Based on Spectral-Spatial Features

Fig. 3. Classification results of different methods in the Pavia University dataset. (a) Color image; (b) classification result of SVM; (c) classification result of CNN; (d) classification result of our method (Log-Gabor and CNN)

Download full size

Fig. 4. Classification results of different method in the Indian Pines dataset. (a) Color image; (b) classification result of SVM; (c) classification result of CNN; (d) classification result of our method (Log-Gabor and CNN)

Download full size

Table 1. Number of training and test samples quantity in the Pavia University dataset
View table
Table 1. Number of training and test samples quantity in the Pavia University dataset
No. Class Number of sample
Training Test
1 Asphalt 65 6566
2 Meadows 185 18464
3 Gravel 20 2079
4 Trees 30 3034
5 Metal sheets 15 1330
6 Bare soil 50 4979
7 Bitumen 15 1315
8 Bricks 35 3647
9 Shadows 10 937

Table 2. Number of training and test samples quantity in the Indian Pines dataset

View table

Table 2. Number of training and test samples quantity in the Indian Pines dataset

No.	Class	Number of sample		No.	Class	Number of sample
No.	Class	Training	Test	No.	Class	Training	Test
1	Alfalfa	10	36	9	Oats	5	15
2	Corn-notill	140	1288	10	Soybean-notill	95	973
3	Corn-mintill	80	750	11	Soybean-mintill	245	2210
4	Corn	25	212	12	Soybean-clean	60	533
5	Grass-pasture	50	432	13	Wheat	20	185
6	Grass-trees	70	660	14	Woods	125	1140
7	Grass-pasture-mowed	10	15	15	Building-grass-trees-drives	40	346
8	Hay-windrowed	50	428	16	Stone-steel-towers	15	78

Table 3. Classification accuracy of different methods in Pavia University dataset
View table
Table 3. Classification accuracy of different methods in Pavia University dataset
Method OA /% AA /% Kappa
SVM 85.96 89.57 0.8940
CNN 96.32 96.91 0.9606
Log-Gabor and CNN 98.41 98.05 0.9833

Table 4. Classification accuracy of different methods in Indian Pines dataset
View table
Table 4. Classification accuracy of different methods in Indian Pines dataset
Method OA /% AA /% Kappa
SVM 76.26 73.29 0.7363
CNN 93.91 80.12 0.9217
Log-Gabor and CNN 95.29 82.65 0.9376

Tools

Get Citation

Copy Citation Text

Qing Fu, Chen Guo, Wenlang Luo. A Hyperspectral Image Classification Method Based on Spectral-Spatial Features[J]. Laser & Optoelectronics Progress, 2020, 57(20): 202803

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Remote Sensing and Sensors

Received: Jan. 16, 2020

Accepted: Mar. 9, 2020

Published Online: Oct. 14, 2020

The Author Email: Guo Chen (fvqing@163.com)

DOI:10.3788/LOP57.202803

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Number of training and test samples quantity in the Pavia University dataset

Table 1. Number of training and test samples quantity in the Pavia University dataset

Table 2. Number of training and test samples quantity in the Indian Pines dataset

Table 2. Number of training and test samples quantity in the Indian Pines dataset

Table 3. Classification accuracy of different methods in Pavia University dataset

Table 3. Classification accuracy of different methods in Pavia University dataset

Table 4. Classification accuracy of different methods in Indian Pines dataset

Table 4. Classification accuracy of different methods in Indian Pines dataset