Cloud-Detection Algorithm for Images Obtained Using the Visual and Infrared Multispectral Imager

Shulin Pang; Lin Sun; Yongming Du; Yanan Tian

doi:10.3788/LOP230439

Laser & Optoelectronics Progress, Volume. 60, Issue 22, 2228003(2023)

Cloud-Detection Algorithm for Images Obtained Using the Visual and Infrared Multispectral Imager

Shulin Pang¹, Lin Sun^1、*, Yongming Du^2、**, and Yanan Tian¹

¹College of Surveying and Spatial Information, Shandong University of Science and Technology, Qingdao 266590, Shandong, China

²State Key Laboratory of Remote Sensing Science, Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(10)

Fig. 1. Principle of threshold determination

Download full size

Fig. 2. Comparison in the results whether adding clear sky restoral test. (a) Clear sky restoral test over water; (b) clear sky restoral test at the sea-land junction; (c) clear sky restoral test using NDVI; (d) clear sky restoral test with ice and snow detection

Download full size

Fig. 3. Comparison of cloud detection results between Fmask and I-Fmask algorithms for different surface types, are the original image, the Fmask result image, the I-Fmask result image, and the true cloud mask.(a) Water; (b) land and water junction; (c) water and vegetation; (d) vegetation; (e) vegetation and mountain; (f) vegetation and urban

Download full size

Fig. 4. Comparison of cloud detection results between Fmask and I-Fmask algorithms for bright surface, are the original image, the Fmask result image, the I-Fmask result image and the true cloud mask. (a) Urban; (b) desert; (c) bare; (d) rock

Download full size

Fig. 5. Comparison of cloud detection results between Fmask and I-Fmask algorithms for mountain area,are the original image, the Fmask result image, the I-Fmask result image and the true cloud mask. (a) Low relief mountain and vegetation; (b) mountain and bare; (c) mountain and snow; (d) high relief mountain and vegetation

Download full size

Fig. 6. Comparison of ice/snow detection results between Fmask and I-Fmask algorithms, from left to right are the original image, the Fmask result image, and the I-Fmask result image. (a) Sea ice; (b)‒(d) snow

Download full size

Table 1. Detailed information about the GF5B VIMI and Landsat 7 TM/ETM+

View table

Table 1. Detailed information about the GF5B VIMI and Landsat 7 TM/ETM+

Landsat 7 TM			Landsat 7 ETM+			GF5B VIMI
Band index	Wavelength range /μm	Spatial resolution /m	Band index	Wavelength range /μm	Spatial resolution /m	Band index	Wavelength range /μm	Spatial resolution /m
1	0.45‒0.52	30	1	0.414‒0.514	30	1	0.45‒0.52	20
2	0.52‒0.60	30	2	0.519‒0.601	30	2	0.52‒0.60	20
3	0.63‒0.69	30	3	0.631‒0.692	30	3	0.62‒0.68	20
4	0.76‒0.90	30	4	0.76‒0.90	30	4	0.76‒0.86	20
5	1.55‒1.75	30	5	1.55‒1.75	30	5	1.55‒1.75	20
7	2.08‒2.35	30	7	2.08‒2.35	3	6	2.08‒2.35	20
						7	3.50‒3.90	40
						8	4.85‒5.05	40
						9	8.01‒8.39	40
						10	8.42‒8.83	40
6	10.4‒12.5	60	6	10.4‒12.5	60	11	10.3‒11.3	40
						12	11.4‒12.5	40

Table 2. Statistical results of cloud amount and accuracy of images with the Fmask algorithm and I-Fmask algorithm
View table
Table 2. Statistical results of cloud amount and accuracy of images with the Fmask algorithm and I-Fmask algorithm
Algorithm R² Slope Intercept MAE /% RMSE /% CAD /% STD
Fmask 0.93 1.001 0.026 4.012 8.024 1.905 0.123
I-Fmask 0.96 0.975 -0.001 2.394 5.846 -2.033 0.095

Table 3. Accuracy evaluation of cloud detection results
View table
Table 3. Accuracy evaluation of cloud detection results
Algorithm R_OA R_CRA R_SRA R_CRM R_SRM
Fmask 0.895 0.851 0.924 0.149 0.076
I-Fmask 0.920 0.924 0.918 0.076 0.082

Table 4. Accuracy and error statistics for cloud detection over diverse land-use types

View table

Table 4. Accuracy and error statistics for cloud detection over diverse land-use types

Surface type	Algorithm	R_CRA	R_SRA	R_CRM	R_SRM	STD
Vegetation	Fmask	0.909	0.910	0.091	0.090	0.028
Vegetation	I-Fmask	0.931	0.941	0.069	0.059	0.023
Water	Fmask	0.960	0.945	0.040	0.055	0.033
Water	I-Fmask	0.975	0.969	0.025	0.031	0.029
Snow	Fmask	0.811	0.941	0.189	0.059	0.160
Snow	I-Fmask	0.850	0.905	0.150	0.095	0.125
Barren/Desert	Fmask	0.917	0.923	0.083	0.077	0.074
Barren/Desert	I-Fmask	0.938	0.906	0.062	0.094	0.057
Urban	Fmask	0.871	0.916	0.129	0.084	0.061
Urban	I-Fmask	0.922	0.882	0.078	0.118	0.043

Tools

Get Citation

Copy Citation Text

Shulin Pang, Lin Sun, Yongming Du, Yanan Tian. Cloud-Detection Algorithm for Images Obtained Using the Visual and Infrared Multispectral Imager[J]. Laser & Optoelectronics Progress, 2023, 60(22): 2228003

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. 3, 2023

Accepted: Mar. 1, 2023

Published Online: Nov. 16, 2023

The Author Email: Lin Sun (sunlin6@126.com), Yongming Du (duym@aircas.ac.cn)

DOI:10.3788/LOP230439

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Detailed information about the GF5B VIMI and Landsat 7 TM/ETM+

Table 1. Detailed information about the GF5B VIMI and Landsat 7 TM/ETM+

Table 2. Statistical results of cloud amount and accuracy of images with the Fmask algorithm and I-Fmask algorithm

Table 2. Statistical results of cloud amount and accuracy of images with the Fmask algorithm and I-Fmask algorithm

Table 3. Accuracy evaluation of cloud detection results

Table 3. Accuracy evaluation of cloud detection results

Table 4. Accuracy and error statistics for cloud detection over diverse land-use types

Table 4. Accuracy and error statistics for cloud detection over diverse land-use types