In this paper, to break the limit of the traditional linear models for synthetic aperture radar (SAR) image despeckling, we propose a novel deep learning approach by learning a non-linear end-to-end mapping between the noisy and clean SAR images with a dilated residual network (SAR-DRN). SAR-DRN is based on dilated convolutions, which can both enlarge the receptive field and maintain the filter size and layer depth with a lightweight structure. In addition, skip connections and a residual learning strategy are added to the despeckling model to maintain the image details and reduce the vanishing gradient problem. Compared with the traditional despeckling methods, the proposed method shows a superior performance over the state-of-the-art methods in both quantitative and visual assessments, especially for strong speckle noise.

Traditionally, clean reference images are needed to train the networks when applying the deep learning techniques to tackle image denoising tasks. However, this idea is impracticable for the task of synthetic aperture radar (SAR) image despeckling, since no real-world speckle-free SAR data exist. To address this issue, this article presents a noisy reference-based SAR deep learning filter, by using complementary images of the same area at different times as the training references. In the proposed method, to better exploit the information of the images, parameter-sharing convolutional neural networks are employed. Furthermore, to mitigate the training errors caused by the land-cover changes between different times, the similarity of each pixel pair between the different images is utilized to optimize the training process. The outstanding despeckling performance of the proposed method was confirmed by the experiments conducted on several multitemporal data sets, when compared with some of the state-of-the-art SAR despeckling techniques. In addition, the proposed method shows a pleasing generalization ability on single-temporal data sets, even though the networks are trained using finite input-reference image pairs at a different imaging area.

Geostationary satellite land surface temperature (GLST) data are important for various dynamic environmental and natural resource applications for terrestrial ecosystems. Due to clouds, shadows, and other atmospheric conditions, the derived LSTs are often missing a large number of values. Reconstructing the missing values is essential for improving the usability of the geostationary satellite LST data. However, current reconstruction methods mainly aim to fill the values of a small number of invalid pixels with many valid pixels, which can provide useful land surface temperature values. When the missing data extent becomes large, the reconstruction effect will worsen because the relationship between different spatiotemporal geostationary satellite LSTs is complex and highly nonlinear. Inspired by the superiority of the deep convolutional neural network (CNN) in solving highly nonlinear and dynamic problems, a multiscale feature connection CNN model is proposed to fill missing LSTs with large missing regions. The proposed method has been tested on both FengYun-2G and Meteosat Second Generation-Spinning Enhanced Visible and InfraRed Imager geostationary satellite LST datasets. The results of simulated and actual experiments show that the proposed method is accurate to within about 1 °C, with 70% missing data rates. This is feasible and effective for large regions of LST reconstruction tasks.

Oil spill accidents can cause severe ecological disasters; hence, the timely and effective detection of oil spills on the marine surface is of great significance. Synthetic aperture radar (SAR) is very suitable for large-scale oil spill monitoring. As a more advanced form of SAR, polarimetric SAR (PolSAR) can provide more scattering information of land objects, which can help to improve the accuracy of oil spill detection. However, the current studies of oil spill detection by SAR data have mainly focused on using SAR intensity or amplitude information, and the phase information and other polarimetric information have not been fully utilized. To solve this problem, using Sentinel-1 dual-polarimetric images as the data source, this article presents an intelligent oil spill detection architecture based on a deep convolutional neural network (DCNN), in which both the amplitude information and phase information are utilized. Furthermore, to improve the feature discrimination capability, the Cloude polarimetric decomposition parameters are also integrated into the proposed model. The results show that the improved DeepLabv3+ model, which takes ResNet-101 as the backbone network and group normalization (GN) as the normalization layer, can achieve superior performance than those traditional methods. Moreover, the model is better able to capture the fine details of oil spill instances and can achieve fine-scale segmentation.

The use of synthetic aperture radar (SAR) technology with quad-polarization data requires efficient polarimetric SAR (PolSAR) speckle filtering algorithms. During the last three decades, many effective methods have been developed to reduce the speckle in PolSAR images, and recent studies have generally shown a trend developing from local single-point filtering to nonlocal patch-based or globally collaborative filtering. The main goals of this paper are to make a comprehensive review of the existing PolSAR despeckling algorithms and highlight the recent development trends. In the experimental part, the filtering results obtained with both simulated and real PolSAR images are deployed to compare the performance of some of the state-of-the-art despeckling algorithms, which shows that all of the selected filters have their individual strengths and weaknesses.