Abstract
Cloud detection, which is defined as the pixel-wise binary classification, is significant in satellite imagery processing. In current remote sensing literature, cloud detection methods are linked to the relationships of imagery bands or based on simple image feature analysis. These methods, which only focus on low-level features, are not robust enough on the images with difficult land covers, for clouds share similar image features such as color and texture with the land covers. To solve the problem, in this paper, we propose a novel deep learning method for cloud detection on satellite imagery by utilizing multilevel image features with two major processes. The first process is to obtain the cloud probability map from the designed deep convolutional neural network, which concatenates deep neural network features from low-level to high-level. The second part of the method is to get refined cloud masks through a composite image filter technique, where the specific filter captures multilevel features of cloud structures and the surroundings of the input imagery. In the experiments, the proposed method achieves 85.38% intersection over union of cloud in the testing set which contains 100 Gaofen-1 wide field of view images and obtains satisfactory visual cloud masks, especially for those hard images. The experimental results show that utilizing multilevel features by the combination of the network with feature concatenation and the particular filter tackles the cloud detection problem with improved cloud masks.
Highlights
We have moved into a Big Data Era [1,2], and an enormous amount of data are expected to be processed to accomplish different special tasks
It is worth noting that Fmask considered almost all the band information with more physical tests conducted such as water test and whiteness test, and cloud shadow detection is carefully designed through the projection analysis, which can be viewed as an extension of the automatic cloud cover assessment (ACCA) method
We propose our cloud detection method with two major processes: the first is FEature Concatenation Network, FECN, a particular type of fully convolutional neural network, and the other is Multi-Window Guided Filtering, MWGF
Summary
We have moved into a Big Data Era [1,2], and an enormous amount of data are expected to be processed to accomplish different special tasks. A modified version of ACCA was developed for Gaofen-1 wide field of view (GF-1 WFV) imagery [15], where Band 2–4 are used in the very first steps to obtain cloud masks (both high confidence and low confidence) and clear sky Another series of physical cloud detection methods are Fmask [16,17,18], which suits Landsat series and Sentinel 2 imagery. MFC Algorithm [19], which utilized the reflectance of all band information, the relationship of bands in GF-1 WFV imagery and analyze the cloud shadow, can be viewed as a type of Fmask algorithm These methods can obtain fine cloud masks, they rely on the reflectance of the imagery bands and the previous threshold setting, which lacks the flexibility and may not be appropriate in difficult situations where there are bright land covers in the imagery, and the reflectance of these land covers is similar to the cloud
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
