文章基本信息

标题：Superresolution Imaging With a Deep Multipath Network for the Reconstruction of Satellite Cloud Images
本地全文：下载
作者：Jinglin Zhang ; Zhipeng Yang ; Zhaoying Jia 等
期刊名称：Earth and Space Science
电子版ISSN：2333-5084
出版年度：2021
卷号：8
期号：6
页码：e2020EA001559
DOI：10.1029/2020EA001559
出版社：John Wiley & Sons, Ltd.
摘要：Satellite cloud images play an important role in weather analysis and forecast. High-resolution satellite images play a significant role in the study of mesoscale weather systems such as typhoons. With the increasing demands of locating and tracking techniques, the resolution of satellite images is no longer satisfactory. Enhancing their resolution with superresolution (SR) methods can help in identifying and locating weather systems. In this paper, we propose a multipath network model, called SRCloudNet, that involves joint training of a back-projection network and a local residual network. SRCloudNet integrates features extracted from back-projection units and residual dense blocks to achieve more accurate image reconstruction. We also developed a novel natural-color cloud and contrail image data set, constituting the first-ever satellite cloud image data set established for SR research. Because of the special features, contrail images were first used to test the performance of SRCloudNet. Extensive experiments demonstrated that SRCloudNet achieves superior performance. Plain Language Abstract Resolution is a crucial indicator for assessing the quality of satellite cloud images. Enhancing the image resolution by improving satellite sensors is straightforward but expensive and difficult to promote. The investigation of satellite cloud image superresolution (SR) is performed using convolutional neural networks (CNNs). Reconstructing images using a traditional SR method causes blurring and loss of high-frequency details. By contrast, ice clouds and water clouds that differ in texture because of their different properties can be effectively distinguished in a natural-color RGB composite image. In this research, a natural-color RGB composite satellite cloud image data set was created to train and test CNNs. Moreover, with consideration of the special features of contrails, a contrail satellite cloud image data set was developed for further testing CNNs. Enhancing the resolution of a contrail image is crucial for calculating contrail coverage and studying the contrail radiation effect. The proposed method, SRCloudNet, can outperform conventional satellite cloud image SR methods.
关键词：contrails;deep learning;Himawari-8 satellite;satellite cloud image;superresolution