Abstract
Abstract. Cloud detection is a vital preprocessing step for remote sensing image applications, which has been widely studied through Convolutional Neural Networks (CNNs) in recent years. However, the available CNN-based works only extract local/non-local features by stacked convolution and pooling layers, ignoring global contextual information of the input scenes. In this paper, a novel segmentation-based network is proposed for cloud detection of remote sensing images. We add a multi-class classification branch to a U-shaped semantic segmentation network. Through the encoder-decoder architecture, pixelwise classification of cloud, shadow and landcover can be obtained. Besides, the multi-class classification branch is built on top of the encoder module to extract global context by identifying what classes exist in the input scene. Linear representation encoded global contextual information is learned in the added branch, which is to be combined with featuremaps of the decoder and can help to selectively strengthen class-related features or weaken class-unrelated features at different scales. The whole network is trained and tested in an end-to-end fashion. Experiments on two Landsat-8 cloud detection datasets show better performance than other deep learning methods, which finally achieves 90.82% overall accuracy and 0.6992 mIoU on the SPARCS dataset, demonstrating the effectiveness of the proposed framework for cloud detection in remote sensing images.
Highlights
Due to their imaging mechanism, optical satellite sensors are inevitably influenced by cloud which can severely degrade image quality or even completely occlude land-covers
The available semantic segmentation networks used for cloud detection only utilize local or non-local features through stacked convolution and pooling layers while ignoring global contextual information of the whole input scene
For large remote sensing images, they have to be cropped into small blocks before being fed into Convolutional Neural Networks (CNNs) because of limited computational resources, which may result in part of cropped blocks having no cloud pixels or no clear pixels at all
Summary
Due to their imaging mechanism, optical satellite sensors are inevitably influenced by cloud which can severely degrade image quality or even completely occlude land-covers. Remote sensing images acquired by such sensors may contaminated by cloud with high probability, hindering their downstream applications such as land-cover classification, change detection, environment monitoring and so on. The rule-based methods exploit reflectance variance in different bands and introduce sets of rules that threshold on single spectral band or combination of spectral bands to identify cloud in remote sensing images. Traditional machine learning methods train classifiers to identify cloud pixels or objects described by handcrafted features. CNNs are trained to classify superpixels or sliding windows of the input image to generate a cloud probability map, cloud detection result is obtained by setting proper threshold on that map. The available semantic segmentation networks used for cloud detection only utilize local or non-local features through stacked convolution and pooling layers while ignoring global contextual information of the whole input scene. Global contextual information about what classes presented in the input scene may help to prevent networks from
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
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 callMore From: ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Disclaimer: 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.
