Cloud detection is a crucial step in the optical satellite image processing pipeline for Earth observation. Clouds in optical remote sensing images seriously affect the visibility of the background and greatly reduce the usability of images for land applications. Traditional methods based on thresholding, multi-temporal or multi-spectral information are often specific to a particular satellite sensor. Convolutional Neural Networks for cloud detection often require labeled cloud masks for training that are very time-consuming and expensive to obtain. To overcome these challenges, this paper presents a hybrid cloud detection method based on the synergistic combination of generative adversarial networks (GAN) and a physics-based cloud distortion model (CDM). The proposed weakly-supervised GAN-CDM method (available online https://github.com/Neooolee/GANCDM) only requires patch-level labels for training, and can produce cloud masks at pixel-level in both training and testing stages. GAN-CDM is trained on a new globally distributed Landsat 8 dataset (WHUL8-CDb, available online doi:https://doi.org/10.5281/zenodo.6420027) including image blocks and corresponding block-level labels. Experimental results show that the proposed GAN-CDM method trained on Landsat 8 image blocks achieves much higher cloud detection accuracy than baseline deep learning-based methods, not only in Landsat 8 images (L8 Biome dataset, 90.20% versus 72.09%) but also in Sentinel-2 images (“S2 Cloud Mask Catalogue” dataset, 92.54% versus 77.00%). This suggests that the proposed method provides accurate cloud detection in Landsat images, has good transferability to Sentinel-2 images, and can quickly be adapted for different optical satellite sensors.

Optical remote sensing has emerged as a crucial technique for earth observation. However, interference of clouds and fog can adversely affect the spatial and spectral information of the images, presenting significant challenges in interpreting remote sensing data and limiting its availability. Moreover, existing methods for addressing the issue of cloud occlusion primarily rely on either physical models or neural networks, lacking a comprehensive integration of the advantages offered by both approaches. So, we propose an end-to-end Asymmetric Unet (AU-Net) dehazing network based on an atmospheric scattering model, which realizes the joint optimization of physical parameters. Firstly, we utilize the Unet to estimate the relatively constant atmospheric light A . Simultaneously, AU-Net based on dark channel prior guidance is used to estimate more complex transmission map T . Next, the atmospheric scattering model is used to deduce the rough dehazing image J based on A and T . Finally, we refine J using AU-Net once again. More specifically, AU-Net incorporates a plug-and-play attention module, namely, Self-Attention with Depth (SAD) and Channel Attention (CA). SAD enhances the fusion of deep semantic information and shallow detail features by expanding the network’s depth, thereby effectively alleviating the dependence of the network on long-distance information. On the other hand, CA addresses the limitations of SAD in terms of channel information. To validate the effectiveness and applicability of proposed dehazing network and attention module, we have achieved state-of-the-art performance and visually appealing results on remote sensing and natural image datasets. The code will be made public on https://github.com/Dudujia160918/Dehazing_Network_AU-Net.

Multi-spectral remote sensing images are widely used for monitoring the globe. Although thin clouds can affect all optical bands, the influences of thin clouds differ with band wavelength. When processing multi-spectral bands at different resolutions, many methods only remove thin clouds in visible/near infrared bands or rescale multi resolution bands to the same resolution then process them together. The former can not make full use of multi-spectral information, and in the latter the rescaling process will introduce noise. In this paper, a deep learning-based thin cloud removal method which fuses full spectral and spatial features in original Sentinel-2 bands is proposed, named CR4S2. A multi input and output architecture is designed for better fusing information in all bands, and reconstructing the background at original resolutions. In addition, two parallel down-sampling residual blocks are designed to transfer features extracted from different depths to the bottom of the network. Experiments were conducted on a new globally distributed Sentinel-2 thin cloud removal dataset called WHUS2-CRv. The results show that the best averaged PSNR, SSIM, nRMSE and SAM of the proposed method over 12 bands in all 20 testing images were 39.55, 0.9443, 0.0245 and 2.5676°, respectively. Compared with baseline methods, the proposed CR4S2 method can better restore not only the spatial features but also spectral features. This indicates that the proposed method is very promising for removing thin clouds in multispectral remote sensing image at different resolutions.

Large area array CCD splicing technology and hardware damage lead to the lack of information in high-resolution remote sensing images (HRSIs), which seriously restricts the application of remote sensing images. However, the complex spatial geographical relationship and the serious spatial heterogeneity will cause texture blur and incomplete structure, especially in the boundary and highly textured regions of HRSI. To deal with the challenge, this article proposes a coarse-to-fine deep generative model with a novel spatial semantic attention (SSA) mechanism. SSA is deployed in the encoder and decoder of the network to ensure the continuity of local features and the relevance of global semantic information and embedded in fine network. In the coarse network, a multilevel loss is proposed to provide more accurate information. Fine network performs more detailed image inpainting operations based on coarse network. Extensive experiments on three datasets, including GF2 satellite images, aerial images, and natural ones, demonstrate that our proposed model has better inpainting effect than the state-of-the-art algorithms.

The accurate estimation and monitoring of phenology is necessary for modern agricultural industries. For crops with short phenology occurrence times, such as rice, Sentinel-1 can be used to effectively monitor the growth status in different phenology periods within a short time interval. Therefore, this study proposes a method to monitor rice phenology based on growth curve simulation by constructing a polarized growth index (PGI) and obtaining a polarized growth curve. A recursive neural network is used to realize the classification of phenology and use it as prior knowledge of rice phenology to divide and extract the phenological interval and date of rice in 2021. The experimental results show that the average accuracy of neural network phenological interval division reaches 93.5%, and the average error between the extracted and measured phenological date is 3.08 days, which proves the application potential of the method. This study will contribute to the technical development of planning, management and maintenance of renewable energy infrastructure related to phenology.

In the UAV path planning study, due to the relative movement of multiple targets and the UAV, long-term and large-scale UAV, autonomous tracking has not been achieved. Therefore, aiming at this problem, this paper uses radar to provide a real-time feedback on target position, estimates the later motion state of the target according to its position, and then perform dynamic path planning by combining the feedback data and the state estimation result. Finally, The UAV path is optimized in real time. Experiments show that the proposed scheme can better plan the UAV path when multiple targets are in motion, thus improving the intelligence of the UAV and the capability of long time tracking.