Cloud Detection Algorithm for Remote Sensing Images Using Fully Convolutional Neural Networks

Mohajerani, Sorour; Krammer, Thomas; Saeedi, Parvaneh

doi:10.48550/arxiv.1810.05782

Cited by 11 publications

(12 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These results prove that the combination of the U-Net architecture and the attention mechanism is effective and that this new neural network architecture performs well in cloud detection tasks. To further validate the proposed neural network architecture, we compare the proposed cloud detection architecture with FCN [51], Fmask [16], the original U-Net architecture and Cloud-Net [48]. Table 2 shows the experimental results of the different methods on the Landsat-Cloud dataset.…”

Section: Resultsmentioning

confidence: 99%

Cloud Detection for Satellite Imagery Using Attention-Based U-Net Convolutional Neural Network

et al. 2020

View full text Add to dashboard Cite

Cloud detection is an important and difficult task in the pre-processing of satellite remote sensing data. The results of traditional cloud detection methods are often unsatisfactory in complex environments or the presence of various noise disturbances. With the rapid development of artificial intelligence technology, deep learning methods have achieved great success in many fields such as image processing, speech recognition, autonomous driving, etc. This study proposes a deep learning model suitable for cloud detection, Cloud-AttU, which is based on a U-Net network and incorporates an attention mechanism. The Cloud-AttU model adopts the symmetric Encoder-Decoder structure, which achieves the fusion of high-level features and low-level features through the skip-connection operation, making the output results contain richer multi-scale information. This symmetrical network structure is concise and stable, significantly enhancing the effect of image segmentation. Based on the characteristics of cloud detection, the model is improved by introducing an attention mechanism that allows model to learn more effective features and distinguish between cloud and non-cloud pixels more accurately. The experimental results show that the method proposed in this paper has a significant accuracy advantage over the traditional cloud detection method. The proposed method is also able to achieve great results in the presence of snow/ice disturbance and other bright non-cloud objects, with strong resistance to disturbance. The Cloud-AttU model proposed in this study has achieved excellent results in the cloud detection tasks, indicating that this symmetric network architecture has great potential for application in satellite image processing and deserves further research.

show abstract

Section: Resultsmentioning

confidence: 99%

Cloud Detection for Satellite Imagery Using Attention-Based U-Net Convolutional Neural Network

et al. 2020

View full text Add to dashboard Cite

show abstract

“…snow and ice in e.g. preprocessing a subtraction from ground truth mask and treatment as distinct classes, in the style of [38], or by searching for an optimal fine-grained architecture and the hyperparameters α, β, and γ using a more systematic validation procedure.…”

Section: F Discussionmentioning

confidence: 99%

“…This still remains a practical challenge despite growing literature, and typical neural network approaches still either train exclusively on small full-resolution patches or on heavily undersampled images. In the context of cloud masking, Shao et al [18] use a CNN for segmenting inputs of 128x128x10 MSI patches, Yang et al [19] apply a CNN for 321x321 RGB or grayscale images obtained by patching a downsampled MSI image, and Moharejani et al [38] used patches of 196x196x4 in combination with QA snow/ice masks. A CNN encoder-decoder of Segal-Rozenheimer et al [39], inspired by DeepLab [40], uses a module of varying-size dilated convolutions before the feature extraction layer, and eventually trains the network on 256x256 patches.…”

Section: Introductionmentioning

confidence: 99%

“…Other remote sensing segmentation applications have benefited from the use of CNNs as well, for instance land cover and crop classification [3]. We are inspired by the properties of Fully Convolutional Networks and the UNet architecture and their descendants, as were other authors [38], [19], [13] that used them recently on remote sensing data. However, we apply CNNs in a setting of approximate mask annotations to an MSI image at full and reduced resolutions.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multi-scale Cloud Detection in Remote Sensing Images using a Dual Convolutional Neural Network

Luotamo,

Metsämäki,

Klami

2020

Preprint

View full text Add to dashboard Cite

Semantic segmentation by convolutional neural networks (CNN) has advanced the state of the art in pixel-level classification of remote sensing images. However, processing large images typically requires analyzing the image in small patches, and hence features that have large spatial extent still cause challenges in tasks such as cloud masking. To support a wider scale of spatial features while simultaneously reducing computational requirements for large satellite images, we propose an architecture of two cascaded CNN model components successively processing undersampled and full resolution images. The first component distinguishes between patches in the inner cloud area from patches at the cloud's boundary region. For the cloudambiguous edge patches requiring further segmentation, the framework then delegates computation to a fine-grained model component. We apply the architecture to a cloud detection dataset of complete Sentinel-2 multispectral images, approximately annotated for minimal false negatives in a land use application. On this specific task and data, we achieve a 16% relative improvement in pixel accuracy over a CNN baseline based on patching.

show abstract

“…In recent years, ML methods have been increasingly applied in the field of atmospheric remote sensing. For example, a neural network (NN) and a convolutional NN are used in the classification of satellite images (Hughes & Hayes, 2014;Mohajerani et al, 2018;Saponaro et al, 2013). Furthermore, an NN can provide a fast and accurate replacement for radiative transfer model (RTM) calculation (Takenaka et al, 2011), especially for hyperspectral and multi-angle simulations, which accelerates the optimal estimation retrieval of atmospheric properties for aerosol and clouds (Chen et al, 2018;Nanda et al, 2019;Segal-Rozenhaimer et al, 2018).…”

mentioning

confidence: 99%

Analytical Prediction of Scattering Properties of Spheroidal Dust Particles With Machine Learning

Chen

Wang

Gomes

et al. 2022

Geophysical Research Letters

View full text Add to dashboard Cite

Dust aerosol has a widespread impact on air quality and visibility (Huang et al., 2008;Moulin et al., 1998) and modulates the Earth's radiation budget (Pachauri et al., 2014) via scattering and absorbing both shortwave and thermal infrared radiation (Xu et al., 2017), leading to large uncertainties in climate projection. The radiative forcing and spectral fingerprints of dust particles depend on their bulk optical properties, including the extinction coefficient, single-scattering albedo, and phase matrix. While these properties can be computed exactly based on the Lorenz-Mie theory for spherical particles (Mishchenko et al., 2002), most mineral dust particles have highly irregular shapes with a wide range of particle sizes, posing significant challenges to both remote sensing and climate modeling (Wang et al., 2003(Wang et al., , 2004.Various techniques have been developed for estimating non-spherical particle properties, such as the discrete dipole approximation (DDA;Draine & Flatau, 1994;Yurkin et al., 2007), the extended boundary condition method (EBCM) as an implementation of the T-matrix (Mishchenko et al., 1997;Mishchenko & Travis, 1994),

show abstract

Cloud Detection Algorithm for Remote Sensing Images Using Fully Convolutional Neural Networks

Cited by 11 publications

References 0 publications

Cloud Detection for Satellite Imagery Using Attention-Based U-Net Convolutional Neural Network

Cloud Detection for Satellite Imagery Using Attention-Based U-Net Convolutional Neural Network

Multi-scale Cloud Detection in Remote Sensing Images using a Dual Convolutional Neural Network

Analytical Prediction of Scattering Properties of Spheroidal Dust Particles With Machine Learning

Contact Info

Product

Resources

About