Airborne multidimensional SAR land cover dataset and fusion classification method

被引：0
作者：
Zheng, Nairong [1 ]
Yang, Zi'an [1 ]
Shi, Xianzheng [1 ]
Yang, Hong [2 ,3 ]
Sun, Yue [2 ]
Wang, Feng [1 ]
机构：
[1] Key Laboratory for Information Science of Electromagnetic Waves (Ministry of Education), School of Information Science and Technology, Fudan University, Shanghai
[2] Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing
[3] School of Electronics and Information, Northwestern Polytechnical University, Xi'an
来源：
National Remote Sensing Bulletin | 2024年 / 28卷 / 09期
关键词：
airborne SAR; AIRMDSAR-Map; deep learning; land cover classification; multi-dimensional; remote sensing; semantic segmentation;
D O I：
10.11834/jrs.20242276
中图分类号：
学科分类号：
摘要：
With the development of Synthetic Aperture Radar (SAR) imaging and deep learning, the use of deep learning to classify land cover in SAR images has received extensive attention and applied research. In this study, a high-resolution airborne multidimensional SAR land cover classification dataset is constructed on the basis of the high-resolution airborne data of the Chinese Aeronautic Remote Sensing System (CARSS) for Earth observation, namely, AIR-MDSAR-Map (Airborne Multidimensional Synthetic Aperture Radar Mapping Dataset). The original data are obtained by CARSS, and the platform is a modified Xinzhou 60 remote sensing aircraft. SAR and optical images are generated in accordance with the standard data production process. After imaging processing, radiometric correction, polarization correction, and geometric correction, the original SAR data are preprocessed to form Single Look Complex (SLC) data, and then geometric processing is used to generate SAR DOM products. After image enhancement, splicing, and rough correction, the raw optical data are preprocessed to generate DSM data, and then semiautomatic filtering is performed to produce DEM. Finally, AIR-MDSAR-Map contains polarization SAR images in bands of C, Ka, L, P, and S and high-resolution optical images in Wanning, Hainan, and Sheyang, Jiangsu, with the spatial resolution ranging from 0.2 m to 1 m depending on the band. AIR-MDSAR-Map divides the land cover into nine categories and generates fine pixel-level labels through a semiautomatic labeling algorithm. In this study, the classical semantic segmentation methods in deep learning, such as UNet, SegNet, DeepLab, and HRNet, are used to verify the classification of AIR-MDSAR-Map. At the same time, we test the classification sensitivity of different band images to all kinds of land cover objects. This dataset includes multidimensional SAR images of the same place and time, which can be used for fusion classification research. In this study, multidimensional SAR data are fused and classified through different fusion strategies; model fusion classifies land cover by selectively fusing the models of each band, and the a priori fusion uses the prior information of the classification results in each band to distinguish land cover on defining the priority of objects. These two fusion methods outperform the single band in the performance of some types of land cover and improve the FWIoU and PA by 10%—15%, the FWIoU reaches 69%, and PA is 81%. AIR-MDSAR-Map can satisfy the research and application requirements of different users and can be used to study the characteristics of the same land cover object with different resolutions, bands, and polarizations. Moreover, it can provide a strong promotion for the development of multidimensional SAR applications. The AIR-MDSAR-Map will be available at the ChinaGEOSS Data Sharing Network (http://www.chinageoss.cn). © 2024 Science Press. All rights reserved.
引用
收藏
页码：2209 / 2222
页数：13
相关论文
共 35 条
  • [1] Addabbo P, Focareta M, Marcuccio S, Votto C, Ullo S L., Land cover classification and monitoring through multisensor image and data combination, 2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), pp. 902-905, (2016)
  • [2] Badrinarayanan V, Kendall A, Cipolla R., SegNet: a deep convolutional encoder-decoder architecture for image segmentation, IEEE Transactions on Pattern Analysis and Machine Intelligence, 39, 12, pp. 2481-2495, (2017)
  • [3] Cen Y, Zhang L F, Zhang X, Wang Y M, Qi W C, Tang S L, Zhang P., Aerial hyperspectral remote sensing classification dataset of Xiongan New Area (Matiwan Village), Journal of Remote Sensing, 24, 11, pp. 1299-1306, (2020)
  • [4] Chen B M, Zhou X P., Explanation of current land use condition classification for national standard of the People’s Republic of China, Journal of Natural Resources, 22, 6, pp. 994-1003, (2007)
  • [5] Chen L C, Papandreou G, Kokkinos I, Murphy K, Yuille A L., DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs, IEEE Transactions on Pattern Analysis and Machine Intelligence, 40, 4, pp. 834-848, (2018)
  • [6] Chen L C, Zhu Y K, Papandreou G, Schroff F, Adam H., Encoder-decoder with atrous separable convolution for semantic image segmentation, 15th European Conference on Computer Vision, pp. 833-851, (2018)
  • [7] Chen Q, Kuang G Y, Li J, Sui L, Li D G., Unsupervised land cover/land use classification using PolSAR imagery based on scattering similarity, IEEE Transactions on Geoscience and Remote Sensing, 51, 3, pp. 1817-1825, (2013)
  • [8] Cloude S R, Pottier E., A review of target decomposition theorems in radar polarimetry, IEEE Transactions on Geoscience and Remote Sensing, 34, 2, pp. 498-518, (1996)
  • [9] Deng J, Dong W, Socher R, Li L J, Li K, Fei-Fei L., ImageNet: a large-scale hierarchical image database, 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp. 248-255, (2009)
  • [10] Dierking W, Busche T., Sea ice monitoring by L-band SAR: an assessment based on literature and comparisons of JERS-1 and ERS-1 imagery, IEEE Transactions on Geoscience and Remote Sensing, 44, 4, pp. 957-970, (2006)
1234