An Efficient Approach to Remove Thick Cloud in VNIR Bands of Multi-Temporal Remote Sensing Images
Abstract
:1. Introduction
2. Methodology
2.1. Theoretical Basis and Approaches for Cloud Removal
2.2. Geometric Correction
2.3. Cloud Detection
2.4. DN Value Estimation for the Cloud Cover Pixels
2.5. Validation of the Approach
2.6. Image Processing Procedure of the Approach
- (1)
- Perform geometric correction of the images T(0), T(−1), T(−2), … and T(j) to make sure that their corresponding pixels are spatially in the same geometric position, respectively.
- (2)
- Perform cloud detection to all the images T(0), T(−1), T(−2), … and T(j) to result in their corresponding cloud-mask images C(0), C(−1), C(−2), … and C(j) in which all the pixels are classified as cloud-free pixels with the values set to 1 and cloud-cover pixels set to 0.
- (3)
- Find a cloudy pixel in the target image. This can be done as the procedures: read the ith pixel of the cloud-mask image C(0,i) to determine if the ith pixel in the target image T(0) is cloud-free or cloud-cover. If C(0,i) = 1, the pixel is clear. Then we proceed to read the next pixel (i + 1) of the cloud-mask image C(0,i + 1) and determine if it is clear or cloudy. We continue this procedure until we find a cloud cover pixel C(0,i).
- (4)
- Find the corresponding cloud-free pixel of C(0,i) in the reference images. This can be done as follows: For j = −1, read C(−1,i) to determine if it is clear or not. If C(−1,i) = 0, it is cloudy and we go to next reference image (j = j − 1) to see if the corresponding pixel C(j,i) is cloudy or not until we obtain a cloud-free pixel C(j,i) that equals to 1.
- (5)
- Find 10 neighboring cloud-free pixels centering the cloud cover pixel C(0,i) in the target image T(0) and its corresponding cloud-free pixel C(j,i) in the reference image T(j). This can be done as follows: set 2 searching windows with sufficient sizes such as 5 × 5, 7 × 7, 9 × 9, …, 301 × 301, … pixels centering respectively C(0,i) and C(j,i) and the maximum window is set to m. Computing the spectral difference SDi,k between the pixel C(j,i) and its 10 neighboring pixels in the band k using the Equation (2) to determine their spectral similarity in the band k and set a threshold value n such as 5, 10, 15, 20, 30, 40, 50 … 100, etc. Enlarge the size of the window until we can get 10 corresponding cloud-free pixels that the SDi,k is less than n in two windows, if still not at the maximum window then gradually enlarge threshold value from 5, 10, 15, …, 50, …, 100, … and continue to search ranging from the minimum window size until you find the 10 available pixels.
- (6)
- Establish the dataset TIk of the target image and the RIk of the reference image for the band k. Using the 10 pixels from the step (5), we go to the reference image T(j) to obtain their pixel values to establish dataset RIk. Using the corresponding location of these 10 pixels in the target image, we can also easily get the dataset TIk.
- (7)
- Establish spectral regression between the two datasets TIk and RIk. We can obtain Equation (5) through the least square method of regression analysis with the pixel values in TIk as the dependent variable and those in RIk as the independent variable.
- (8)
- Estimate the DN value of the cloud cover pixel in the target image. We can get DNq,k in Equation (6) using the DN value corresponding to the cloud-free pixel C(i,j) = 1 in reference image T(j) and Equation (6) for band k.
- (9)
- Estimate the DN values of the cloud cover pixel in the target image for other bands. Continue the above steps (5)–(8) until all the bands have been completed.
- (10)
- Repeat the above steps (3)–(9) until all the cloudy pixels in the target image have been removed with their DN values estimated for all the bands.
3. Result and Analysis
3.1. Accuracy of Cloud Removal over Various Ground Surface Patterns
3.2. Impact of Reference Images on Accuracy of Cloud Removal
3.3. Effect of Cloud Cover Sizes on Accuracy of Cloud Removal
3.4. Application to Southeast China Region
4. Discussion
5. Conclusions
Author Contributions
Funding
Acknowledgments
Conflicts of Interest
References
- Duan, S.B.; Li, Z.L.; Tang, B.H.; Wu, H.; Tang, R.L. Generation of a time-consistent land surface temperature product from MODIS data. Remote Sens. Environ. 2014, 140, 339–349. [Google Scholar] [CrossRef]
- Liang, L.; Zhao, S.H.; Qin, Z.H.; He, K.X.; Chong, C.; Luo, Y.X.; Zhou, X.D. Drought change trend using MODIS TVDI and its relationship with climate factors in china from 2001 to 2010. J. Integr. Agric. 2014, 13, 1501–1508. [Google Scholar] [CrossRef]
- Li, Z.L.; Becker, F. Feasibility of land surface temperature and emissivity determination from AVHRR data. Remote Sens. Environ. 1993, 43, 67–85. [Google Scholar] [CrossRef]
- Duan, S.B.; Li, Z.L.; Leng, P. A framework for the retrieval of all-weather land surface temperature at a high spatial resolution from polar-orbiting thermal infrared and passive microwave data. Remote Sens. Environ. 2017, 195, 107–117. [Google Scholar] [CrossRef]
- Rozenstein, O.; Qin, Z.; Derimian, Y.; Karnieli, A. Derivation of land surface temperature for landsat-8 tirs using a split window algorithm. Sensors 2014, 14, 5768–5780. [Google Scholar] [CrossRef]
- Ren, H.; Ye, X.; Liu, R.; Dong, J.; Qin, Q. Improving land surface temperature and emissivity retrieval from the Chinese gaofen-5 satellite using a hybrid algorithm. IEEE Trans. Geosci. Remote Sens. 2017, 56, 1080–1090. [Google Scholar] [CrossRef]
- Gao, M.F.; Qin, Z.H.; Zhang, H.O.; Lu, L.P.; Zhou, X.; Yang, X.C. Remote sensing of agro-droughts in Guangdong province of China using MODIS satellite data. Sensors 2008, 8, 4687–4708. [Google Scholar] [CrossRef] [PubMed]
- Leng, P.; Song, X.N.; Duan, S.B.; Li, Z.L. A practical algorithm for estimating surface soil moisture using combined optical and thermal infrared data. Int. J. Appl. Earth. Obs. 2016, 52, 338–348. [Google Scholar] [CrossRef]
- Tang, R.L.; Li, Z.L.; Tang, B.H. An application of the Ts-VI triangle method with enhanced edges determination for evapotranspiration estimation from MODIS data in arid and semi-arid regions: Implementation and validation. Remote Sens. Environ. 2010, 114, 540–551. [Google Scholar] [CrossRef]
- Tang, B.H.; Li, Z.L.; Zhang, R.H. A direct method for estimating net surface shortwave radiation from MODIS data. Remote Sens. Environ. 2006, 103, 115–126. [Google Scholar] [CrossRef]
- Goodwin, N.R.; Collett, L.J.; Denham, R.J.; Flood, N.; Tindall, D. Cloud and cloud shadow screening across Queensland, Australia: An automated method for Landsat TM/ETM+ time series. Remote Sens. Environ. 2013, 134, 50–65. [Google Scholar] [CrossRef]
- Zhu, Z.; Woodcock, C.E. Object-based cloud and cloud shadow detection in Landsat imagery. Remote Sens. Environ. 2012, 118, 83–94. [Google Scholar] [CrossRef]
- Qiu, S.; He, B.; Zhu, Z.; Liao, Z.; Quan, X. Improving Fmask cloud and cloud shadow detection in mountainous area for Landsats 4-8 images. Remote Sens. Environ. 2017, 199, 107–119. [Google Scholar] [CrossRef]
- Xu, M.; Jia, X.P.; Pickering, M.; Plaza, A.J. Cloud removal based on sparse representation via multitemporal dictionary learning. IEEE Trans. Geosci. Remote Sens. 2016, 54, 2998–3006. [Google Scholar] [CrossRef]
- Chang, N.B.; Bai, K.; Chen, C.F. Smart information reconstruction via time-space-spectrum continuum for cloud removal in satellite images. IEEE J. STARS 2015, 8, 1898–1912. [Google Scholar] [CrossRef]
- Hagolle, O.; Huc, M.; Pascual, D.V.; Dedieu, G. A multi-temporal method for cloud detection, applied to FORMOSAT-2, VENμS, LANDSAT and SENTINEL-2 images. Remote Sens. Environ. 2010, 114, 1747–1755. [Google Scholar] [CrossRef]
- Jin, S.; Homer, C.; Yang, L.; Xian, G.; Fry, J.; Danielson, P.; Townsend, P.A. Automated cloud and shadow detection and filling using two-date Landsat imagery in the USA. Int. J. Remote Sens. 2013, 34, 1540–1560. [Google Scholar] [CrossRef]
- Roy, D.P.; Ju, J.; Kline, K.; Scaramuzza, P.L.; Kovalskyy, V.; Hansen, M.; Loveland, T.R.; Vermote, E.; Zhang, C.S. Web-enabled Landsat data (weld): Landsat ETM+ composited mosaics of the conterminous United States. Remote Sens. Environ. 2010, 114, 35–49. [Google Scholar] [CrossRef]
- Wilson, M.J.; Oreopoulos, L. Enhancing a simple MODIS cloud mask algorithm for the Landsat data continuity mission. IEEE Trans. Geosci. Remote Sens. 2013, 51, 723–731. [Google Scholar] [CrossRef]
- Zhu, Z.; Wang, S.; Woodcock, C.E. Improvement and expansion of the fmask algorithm: Cloud, cloud shadow, and snow detection for Landsats 4-7, 8, and sentinel 2 images. Remote Sens. Environ. 2015, 159, 269–277. [Google Scholar] [CrossRef]
- Shen, H.F.; Li, X.H.; Cheng, Q.; Zeng, C.; Yang, G.; Li, H.F.; Zhang, L.P. Missing information reconstruction of remote sensing data: A technical review. IEEE Geosci. Remote Sens. Mag. 2015, 3, 61–85. [Google Scholar] [CrossRef]
- Shen, H.F.; Li, H.F.; Qian, Y.; Zhang, L.P.; Yuan, Q.Q. An effective thin cloud removal procedure for visible remote sensing images. ISPRS J. Photogramm. 2014, 96, 224–235. [Google Scholar] [CrossRef]
- Surya, S.R.; Simon, P. Automatic cloud removal from multitemporal satellite images. J. Indian Soc. Remote 2014, 43, 57–68. [Google Scholar] [CrossRef]
- Chen, B.; Huang, B.; Chen, L.F.; Xu, B. Spatially and temporally weighted regression: A novel method to produce continuous cloud-free Landsat imagery. IEEE Trans. Geosci. Remote Sens. 2016, 55, 27–37. [Google Scholar] [CrossRef]
- Maalouf, A.; Carre, P.; Augereau, B.; Maloigne, C.F. A bandelet-based inpainting technique for clouds removal from remotely sensed images. IEEE Trans. Geosci. Remote Sens. 2009, 47, 2363–2371. [Google Scholar] [CrossRef]
- Zhu, X.; Gao, F.; Liu, D.; Chen, J. A modified neighborhood similar pixel interpolator approach for removing thick clouds in Landsat images. IEEE Geosci. Sens. Lett. 2012, 9, 521–525. [Google Scholar] [CrossRef]
- Melgani, F. Contextual reconstruction of cloud-contaminated multitemporal multispectral images. IEEE Trans. Geosci. Remote Sens. 2006, 44, 442–455. [Google Scholar] [CrossRef]
- Eckardt, R.; Berger, C.; Thiel, C.; Schmullius, C. Removal of optically thick clouds from multi-spectral satellite images using multi-frequency SAR data. Remote Sens. 2013, 5, 2973–3006. [Google Scholar] [CrossRef]
- Huang, B.; Li, Y.; Han, X.Y.; Cui, Y.Z.; Li, W.B.; Li, R.R. Cloud removal from optical satellite imagery with SAR imagery using sparse representation. IEEE Geosci. Remote Sens. Lett. 2017, 12, 1046–1050. [Google Scholar] [CrossRef]
- Xu, M.; Pickering, M.; Plaza, A.J.; Jia, X.P. Thin cloud removal based on signal transmission principles and spectral mixture analysis. IEEE Trans. Geosci. Remote Sens. 2016, 54, 1659–1669. [Google Scholar] [CrossRef]
- Li, X.H.; Shen, H.F.; Zhang, L.P.; Zhang, H.Y.; Yuan, Q.Q.; Yang, G. Recovering quantitative remote sensing products contaminated by thick clouds and shadows using multitemporal dictionary learning. IEEE Trans. Geosci. Remote Sens. 2014, 52, 7086–7098. [Google Scholar]
- Lin, C.H.; Tsai, P.H.; Lai, K.H.; Chen, J.Y. Cloud removal from multitemporal satellite images using information cloning. IEEE Trans. Geosci. Remote Sens. 2013, 51, 232–241. [Google Scholar] [CrossRef]
- Valero, S.; Pelletier, C.; Bertolino, M. Patch-based reconstruction of high resolution satellite image time series with missing values using spatial, spectral and temporal similarities. In Proceedings of the 2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Beijing, China, 10–15 July 2016; pp. 2308–2311. [Google Scholar]
- Cheng, Q.; Shen, H.F.; Zhang, L.P.; Yuan, Q.Q.; Zeng, C. Cloud removal for remotely sensed images by similar pixel replacement guided with a spatio-temporal MRF model. ISPRS J. Photogramm. 2014, 92, 54–68. [Google Scholar] [CrossRef]
- Foga, S.; Scaramuzza, P.L.; Guo, S.; Zhu, Z.; Jr, R.D.D.; Beckmann, T.; Schmidt, G.L.; Dwyer, J.L.; Hughes, M.J.; Laue, B. Cloud detection algorithm comparison and validation for operational Landsat data products. Remote Sens. Environ. 2017, 194, 379–390. [Google Scholar] [CrossRef] [Green Version]
- Safont, G.; Salazar, A.; Vergara, L.; Rodríguez, A. Nonlinear estimators from ICA mixture models. Signal Process. 2019, 155, 281–286. [Google Scholar] [CrossRef]
Surface Patterns | Methods | RMSE | W | ||||||
---|---|---|---|---|---|---|---|---|---|
Band1 | Band2 | Band3 | Band4 | Band1 | Band2 | Band3 | Band4 | ||
Water | DR | 7.61 | 4.70 | 3.77 | 6.47 | 0.93 | 0.91 | 0.93 | 0.81 |
MSD | 2.28 | 1.09 | 1.58 | 3.46 | 0.98 | 0.98 | 0.97 | 0.90 | |
LRM | 1.25 | 0.86 | 0.91 | 0.68 | 0.99 | 0.98 | 0.98 | 0.98 | |
Plain | DR | 7.54 | 4.97 | 8.36 | 7.41 | 0.93 | 0.90 | 0.86 | 0.91 |
MSD | 2.01 | 1.34 | 2.49 | 3.24 | 0.98 | 0.97 | 0.96 | 0.96 | |
LRM | 1.87 | 1.25 | 2.38 | 3.07 | 0.98 | 0.98 | 0.96 | 0.96 | |
Mountain | DR | 8.98 | 4.33 | 6.58 | 5.73 | 0.90 | 0.89 | 0.83 | 0.94 |
MSD | 1.90 | 1.31 | 2.50 | 5.53 | 0.98 | 0.97 | 0.94 | 0.94 | |
LRM | 1.69 | 0.99 | 0.98 | 1.92 | 0.98 | 0.98 | 0.97 | 0.98 | |
City | DR | 7.60 | 5.44 | 7.03 | 6.64 | 0.93 | 0.90 | 0.88 | 0.89 |
MSD | 1.97 | 1.22 | 1.80 | 1.89 | 0.98 | 0.98 | 0.97 | 0.97 | |
LRM | 1.92 | 1.12 | 1.74 | 1.73 | 0.98 | 0.98 | 0.97 | 0.97 |
Surface Patterns | DOY | Methods | RMSE | W | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Band1 | Band2 | Band3 | Band4 | Band1 | Band2 | Band3 | Band4 | |||
City | DOY124 | DR | 8.14 | 5.56 | 7.22 | 7.02 | 0.93 | 0.90 | 0.88 | 0.89 |
MSD | 2.37 | 1.32 | 1.86 | 1.82 | 0.98 | 0.98 | 0.97 | 0.97 | ||
LRM | 2.31 | 1.29 | 1.62 | 1.78 | 0.98 | 0.98 | 0.97 | 0.97 | ||
DOY108 | DR | 19.50 | 11.01 | 13.61 | 13.58 | 0.83 | 0.79 | 0.78 | 0.78 | |
MSD | 2.90 | 1.84 | 2.38 | 2.90 | 0.97 | 0.97 | 0.96 | 0.95 | ||
LRM | 2.02 | 1.36 | 1.75 | 2.42 | 0.98 | 0.97 | 0.97 | 0.96 | ||
DOY60 | DR | 33.34 | 19.19 | 24.25 | 29.75 | 0.71 | 0.64 | 0.60 | 0.52 | |
MSD | 5.20 | 3.00 | 5.82 | 5.48 | 0.95 | 0.94 | 0.90 | 0.91 | ||
LRM | 2.97 | 1.90 | 2.95 | 3.84 | 0.97 | 0.96 | 0.95 | 0.94 | ||
Water | DOY124 | DR | 7.39 | 4.43 | 3.75 | 6.74 | 0.94 | 0.92 | 0.94 | 0.89 |
MSD | 2.24 | 1.28 | 1.67 | 3.07 | 0.98 | 0.98 | 0.97 | 0.95 | ||
LRM | 1.41 | 1.02 | 0.93 | 0.87 | 0.99 | 0.98 | 0.98 | 0.99 | ||
DOY108 | DR | 17.19 | 8.39 | 8.92 | 10.98 | 0.85 | 0.84 | 0.85 | 0.82 | |
MSD | 2.52 | 1.86 | 1.86 | 6.96 | 0.98 | 0.97 | 0.97 | 0.89 | ||
LRM | 1.57 | 0.95 | 0.97 | 0.95 | 0.99 | 0.97 | 0.98 | 0.92 | ||
DOY60 | DR | 24.14 | 14.59 | 16.61 | 11.74 | 0.79 | 0.73 | 0.73 | 0.81 | |
MSD | 3.53 | 2.43 | 2.79 | 8.57 | 0.97 | 0.95 | 0.95 | 0.86 | ||
LRM | 1.42 | 1.02 | 1.01 | 0.96 | 0.99 | 0.98 | 0.98 | 0.97 |
Quality Index | Band | 5PIX-DOY124 | 10PIX-DOY124 | 20PIX-DOY124 | ||||||
---|---|---|---|---|---|---|---|---|---|---|
DR | MSD | LRM | DR | MSD | LRM | DR | MSD | LRM | ||
RMSE | Band1 | 8.56 | 1.55 | 1.48 | 8.98 | 1.90 | 1.69 | 9.55 | 1.87 | 1.67 |
Band2 | 4.04 | 1.30 | 0.80 | 4.33 | 1.31 | 0.99 | 4.90 | 1.25 | 1.01 | |
Band3 | 6.40 | 1.96 | 1.00 | 6.58 | 2.50 | 0.98 | 7.29 | 2.13 | 1.22 | |
Band4 | 4.61 | 5.23 | 2.23 | 5.73 | 5.53 | 1.92 | 6.87 | 5.79 | 3.26 | |
W | Band1 | 0.91 | 0.98 | 0.98 | 0.90 | 0.98 | 0.98 | 0.90 | 0.98 | 0.98 |
Band2 | 0.90 | 0.97 | 0.98 | 0.89 | 0.97 | 0.98 | 0.88 | 0.97 | 0.97 | |
Band3 | 0.84 | 0.95 | 0.97 | 0.83 | 0.94 | 0.97 | 0.81 | 0.95 | 0.97 | |
Band4 | 0.95 | 0.95 | 0.98 | 0.94 | 0.94 | 0.98 | 0.93 | 0.94 | 0.96 | |
Quality Index | Band | 50PIX-DOY124 | 75PIX-DOY124 | 100PIX-DOY124 | ||||||
DR | MSD | LRM | DR | MSD | LRM | DR | MSD | LRM | ||
RMSE | Band1 | 9.33 | 1.81 | 1.69 | 9.28 | 1.85 | 1.59 | 9.20 | 2.00 | 1.57 |
Band2 | 4.76 | 1.31 | 0.98 | 4.87 | 1.32 | 0.93 | 4.89 | 1.42 | 0.91 | |
Band3 | 7.25 | 2.01 | 1.42 | 7.28 | 2.10 | 1.20 | 7.32 | 2.41 | 1.13 | |
Band4 | 6.24 | 6.01 | 3.02 | 6.02 | 6.65 | 3.17 | 6.19 | 6.92 | 3.16 | |
W | Band1 | 0.90 | 0.98 | 0.98 | 0.90 | 0.98 | 0.98 | 0.90 | 0.98 | 0.98 |
Band2 | 0.88 | 0.97 | 0.98 | 0.88 | 0.97 | 0.98 | 0.88 | 0.97 | 0.98 | |
Band3 | 0.81 | 0.95 | 0.96 | 0.81 | 0.95 | 0.97 | 0.82 | 0.94 | 0.97 | |
Band4 | 0.94 | 0.94 | 0.97 | 0.94 | 0.93 | 0.97 | 0.94 | 0.93 | 0.97 |
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Share and Cite
Du, W.; Qin, Z.; Fan, J.; Gao, M.; Wang, F.; Abbasi, B. An Efficient Approach to Remove Thick Cloud in VNIR Bands of Multi-Temporal Remote Sensing Images. Remote Sens. 2019, 11, 1284. https://doi.org/10.3390/rs11111284
Du W, Qin Z, Fan J, Gao M, Wang F, Abbasi B. An Efficient Approach to Remove Thick Cloud in VNIR Bands of Multi-Temporal Remote Sensing Images. Remote Sensing. 2019; 11(11):1284. https://doi.org/10.3390/rs11111284
Chicago/Turabian StyleDu, Wenhui, Zhihao Qin, Jinlong Fan, Maofang Gao, Fei Wang, and Bilawal Abbasi. 2019. "An Efficient Approach to Remove Thick Cloud in VNIR Bands of Multi-Temporal Remote Sensing Images" Remote Sensing 11, no. 11: 1284. https://doi.org/10.3390/rs11111284