Cloud Detection for GF-5 Visible-Shortwave Infrared Advanced Hyperspectral Image

Jian Wang; Tianxiang Cui; Yi Wang; Lin Sun

doi:10.3788/AOS202141.0928003

Acta Optica Sinica, Volume. 41, Issue 9, 0928003(2021)

Cloud Detection for GF-5 Visible-Shortwave Infrared Advanced Hyperspectral Image

Jian Wang¹, Tianxiang Cui², Yi Wang³, and Lin Sun^4、*

Author Affiliations

¹College of Information and Management Science, Henan Agricultural University, Zhengzhou, Henan 450046, China

²College of Forestry, Nanjing Forestry University, Nanjing, Jiangsu 210042, China

³Geovis Technology Co., Ltd., Beijing 101399, China

⁴College of Geodesy and Geomatics, Shandong University of Science and Technology, Qingdao, Shandong 266590, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(9)

Fig. 1. Hyperspectral images of cloud detection areas (false color composite image of bands 110, 64, 44 in RGB, the red vector is the visually interpreted cloud boundary). (a) Image a; (b) image b; (c) image c; (d) image d

Download full size

Fig. 2. Apparent reflectance curves of different ground objects in AHSI hyperspectral image (the averages are shown as solid dots, and the error bars correspond to the standard deviations)

Download full size

Fig. 3. Band ratio between different types ground objects in AHSI hyperspectral image. (a) Thin cloud; (b) soil; (c) rock

Download full size

Fig. 4. Cloud detection algorithm flow chart

Download full size

Fig. 5. Cloud detection results, white represents cloud, and the red vector is the visually interpreted cloud boundary. (a) Image a; (b) image b; (c) image c; (d) image d

Download full size

Fig. 6. Local comparison of cloud detection results. (a1)(a2) False color image (using bands 110, 64, and 44 to represent the R, G, and B bands, respectively); (b) visual interpretation results, red labeled as cloud pixels; (c) detection results of our algorithm, green labeled as cloud pixels

Download full size

Table 1. Comparison of parameters of GF-5 AHSI and EO-1 Hyperion

View table

Table 1. Comparison of parameters of GF-5 AHSI and EO-1 Hyperion

Instrument	Channel	Spectralrange /μm	Number ofbands	Bandwidth /nm	Swathwidth /km	Spatialresolution /m
GF-5 AHSI	VIS-NIR	0.39-1.004	150	5
	SWIR	1.01-2.51	180	10	60	30
EO-1 Hyperion	VIS-NIR	0.356-1.085	70	10
	SWIR	0.852-2.577	172	10	7.7	30

Table 2. Cloud detection image parameters
View table
Table 2. Cloud detection image parameters
Sensor Image Date Path /row Longitude /(°) Latitude /(°) Solar zenith /(°)
a 2019-04-08 363/530 88.8 44.8 39.35
AHSI b 2019-04-04 377/624 83.2 41.6 38.03
c 2019-04-04 377/623 83.3 41.2 37.61
d 2019-04-08 363/631 88.7 45.1 39.59

Table 3. Statistics of cloud detection accuracy of our algorithm
View table
Table 3. Statistics of cloud detection accuracy of our algorithm
Image Proportion /% Overallaccuracy /% Producer'saccuracy /% User'saccuracy /%
Visual interpretation Our algorithm
a 13.0 10.0 95 71 93
b 42.0 46.7 89 92 83
c 37.6 40.5 93 94 88
d 35.7 30.3 90 79 93

Tools

Get Citation

Copy Citation Text

Jian Wang, Tianxiang Cui, Yi Wang, Lin Sun. Cloud Detection for GF-5 Visible-Shortwave Infrared Advanced Hyperspectral Image[J]. Acta Optica Sinica, 2021, 41(9): 0928003

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites