基于深度学习和Sentinel-1卫星影像的北极海冰分类精度和影响因素

邵志远; 赵杰臣; 解龙翔; 牟芳如; 肖静; 刘敏君; 陈雪婧

doi:10.3969/j.issn.1001-909X.2024.03.010

PDF(4482 KB)

海洋学研究 ›› 2024, Vol. 42 ›› Issue (3) : 119-130. DOI: 10.3969/j.issn.1001-909X.2024.03.010

研究论文

基于深度学习和Sentinel-1卫星影像的北极海冰分类精度和影响因素

邵志远 ¹^,²^,³ ,
赵杰臣 ¹^,²^,⁴^,^* ,
解龙翔 ¹^,²^,³ ,
牟芳如 ¹^,² ,
肖静 ¹^,² ,
刘敏君 ¹^,² ,
陈雪婧 ¹^,²

作者信息 +

Classification accuracy and influencing factors of Arctic sea ice based on deep learning and Sentinel-1 satellite imagery

SHAO Zhiyuan ¹^,²^,³ ,
ZHAO Jiechen ¹^,²^,⁴^,^* ,
XIE Longxiang ¹^,²^,³ ,
MU Fangru ¹^,² ,
XIAO Jing ¹^,² ,
LIU Minjun ¹^,² ,
CHEN Xuejing ¹^,²

Author information +

文章历史 +

摘要

海冰类型是极地海冰的重要属性之一,多年冰的物理性质较一年冰有着显著差异,因此识别海冰类型对极地气候变化研究和冰区船舶航行保障意义重大。卫星遥感是获取多时序、大范围海冰信息的有效手段。该文以北极西北航道和东北航道为研究区域,基于3个深度学习模型(ResNet、Vision Transformer、Swin Transformer)对Sentinel-1卫星双极化合成孔径雷达影像进行海冰分类研究。结果表明,8×8像素切片数据集的海冰分类效果优于其他尺寸切片数据集;对假彩色合成图像进行偏移量处理能够有效地减少噪声对海冰分类的影响;在3个深度学习模型中,Swin Transformer模型分类精度最高,整体准确率和Kappa系数均在98%以上。比较多年冰密集度数据发现,3个模型的结果与AMSR2的偏差均小于10%。

Abstract

The type of sea ice is one of the important attributes of polar sea ice, and the physical properties of multi-year ice are significantly different from those of first-year ice. Therefore, the identifying the types of sea ice is of great significance to the research of polar climate change and the navigation security of ships in ice-covered regions. Satellite remote sensing is an effective method to obtain multi-temporal and large-scale sea ice type information. Based on three deep learning models (ResNet, Vision Transformer, Swin Transformer) and Sentinel-1 satellite dual-polarization synthetic aperture radar (SAR) images, this paper studies the classification method for sea ice in the regions of the Northwest and Northeast Passage in the Arctic. The results showed that the sea ice classification effect of 8×8 pixel slice dataset was better than that of other size slice datasets. Offset processing false color images could effectively reduce the influence of noise on sea ice classification. Among the three deep learning models, the Swin Transformer model had the highest classification accuracy, with the overall accuracy and Kappa coefficient above 98%. Comparing the multi-year ice concentration, it was found that the results of the three models deviate less than 10% from the AMSR2 data.

导出引用

邵志远, 赵杰臣, 解龙翔, 等. 基于深度学习和Sentinel-1卫星影像的北极海冰分类精度和影响因素[J]. 海洋学研究. 2024, 42(3): 119-130 https://doi.org/10.3969/j.issn.1001-909X.2024.03.010

SHAO Zhiyuan, ZHAO Jiechen, XIE Longxiang, et al. Classification accuracy and influencing factors of Arctic sea ice based on deep learning and Sentinel-1 satellite imagery[J]. Journal of Marine Sciences. 2024, 42(3): 119-130 https://doi.org/10.3969/j.issn.1001-909X.2024.03.010

中图分类号： P731.15;P714

参考文献

列表( 原文顺序 | 文献年度倒序 | 文中引用次数倒序 ) 可视化分析

[1]	KIM J W, KIM D J, HWANG B J. Characterization of Arctic sea ice thickness using high-resolution spaceborne polarimetric SAR data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2012, 50(1): 13-22. 本文引用 [1]

[2]	HOSHINO S, TATEYAMA K, IZUMIYAMA K. Classification of ice in Lützow-Holm Bay, East Antarctica, using data from ASCAT and AMSR2[J]. Remote Sensing, 2020, 12(19):3179. 本文引用 [1]

[3]	张晰. 极化SAR渤海海冰厚度探测研究[D]. 青岛: 中国海洋大学, 2011. ZHANG X. Research on sea ice thickness detection by polarimetric SAR in Bohai Sea[D]. Qingdao: Ocean University of China, 2011. 本文引用 [1]

[4]	FARRELL S L, DUNCAN K, BUCKLEY E M, et al. Mapping sea ice surface topography in high fidelity with ICESat-2[J]. Geophysical Research Letters, 2020, 47(21): e2020GL090708. 本文引用 [1]

[5]	CAO X W, LU P, LEI R B, et al. Physical and optical characteristics of sea ice in the Pacific Arctic Sector during the summer of 2018[J]. Acta Oceanologica Sinica, 2020, 39(9): 25-37. 本文引用 [1]

[6]	DIERKING W, LANG O, BUSCHE T. Sea ice local surface topography from single-pass satellite InSAR measurements: A feasibility study[J]. The Cryosphere, 2017, 11(4): 1967-1985. 本文引用 [1]

[7]	RHEINLAENDER J W, DAVY R, ÓLASON E, et al. Driving mechanisms of an extreme winter sea ice breakup event in the Beaufort Sea[J]. Geophysical Research Letters, 2022, 49(12): e2022GL099024. 本文引用 [1]

[8]	刘泉宏, 张韧, 汪杨骏, 等. 深度学习方法在北极海冰预报中的应用[J]. 大气科学学报, 2022, 45(1):14-21. LIU Q H, ZHANG R, WANG Y J, et al. Application of deep learning methods to Arctic Sea ice prediction[J]. Transactions of Atmospheric Sciences, 2022, 45(1): 14-21. 本文引用 [1]

[9]	刘启明, 杨树国, 赵莉. 基于深度卷积神经网络的海洋多目标涡旋检测方法[J]. 青岛科技大学学报:自然科学版, 2022, 43(4):120-126. LIU Q M, YANG S G, ZHAO L. Ocean multi-eddy detection method based on deep convolution neural network[J]. Journal of Qingdao University of Science and Technology: Natural Science Edition, 2022, 43(4): 120-126. 本文引用 [1]

[10]	崔艳荣. 卷积神经网络在渤海海冰卫星遥感中的应用[D]. 上海: 上海海洋大学, 2020. CUI Y R. Application of convolutional neural network in remote sensing of sea ice satellites in the Bohai Sea[D]. Shanghai: Shanghai Ocean University, 2020. 本文引用 [3]

[11]	KHALEGHIAN S, ULLAH H, KRAEMER T, et al. Deep semisupervised teacher-student model based on label propagation for sea ice classification[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14: 10761-10772. 本文引用 [3]

[12]	王芳. 基于Sentinel-1极化数据北冰洋海冰分类研究[D]. 北京: 中国地质大学, 2021. WANG F. Arctic sea ice classification based on Sentinel-1 polarization data[D]. Beijing: China University of Geos-ciences, 2021. 本文引用 [2]

[13]	HAN Y L, GAO Y, ZHANG Y, et al. Hyperspectral sea ice image classification based on the spectral-spatial-joint feature with deep learning[J]. Remote Sensing, 2019, 11(18): 2170. 本文引用 [1]

[14]	张赛, 樊博文, 禹定峰, 等. 基于多尺度融合网络的辽东湾海冰提取方法研究[J]. 海洋测绘, 2023, 43(1):68-72. ZHANG S, FAN B W, YU D F, et al. Research on sea ice extraction method of Liaodong Bay based on multi-scale fusion network[J]. Hydrographic Surveying and Charting, 2023, 43(1): 68-72. 本文引用 [1]

[15]	李金鑫. 基于卷积神经网络的SAR图像分类[D]. 长春: 吉林大学, 2018. LI J X. SAR image classification based on convolutional neural network[D]. Changchun: Jilin University, 2018. 本文引用 [1]

[16]	HAN Y L, WEI C, ZHOU R Y, et al. Combining 3D-CNN and squeeze-and-excitation networks for remote sensing sea ice image classification[J]. Mathematical Problems in Engineering, 2020: 8065396. 本文引用 [1]

[17]	HAN Y L, LIU Y K, HONG Z H, et al. Sea ice image classification based on heterogeneous data fusion and deep learning[J]. Remote Sensing, 2021, 13(4): 592. 本文引用 [1]

[18]	PETROU Z I, TIAN Y L. Prediction of sea ice motion with convolutional long short-term memory networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(9): 6865-6876. 本文引用 [1]

[19]	焦艳, 黄菲, 高松, 等. 基于长短时记忆神经网络的辽东湾海冰延伸期预报方法研究[J]. 中国海洋大学学报:自然科学版, 2020, 50(6):1-11. JIAO Y, HUANG F, GAO S, et al. Research on extended-range forecast model of sea ice in the Liaodong Bay based on long short term memory network[J]. Periodical of Ocean University of China, 2020, 50(6): 1-11. 本文引用 [1]

[20]	LIU Q H, ZHANG R, WANG Y J, et al. Daily prediction of the Arctic sea ice concentration using reanalysis data based on a convolutional LSTM network[J]. Journal of Marine Science and Engineering, 2021, 9(3): 330. 本文引用 [1]

[21]	ZHENG Q Y, LI W, SHAO Q, et al. A mid- and long-term Arctic sea ice concentration prediction model based on deep learning technology[J]. Remote Sensing, 2022, 14(12): 2889. 本文引用 [1]

[22]	WEI J F, HANG R L, LUO J J. Prediction of Pan-Arctic sea ice using attention-based LSTM neural networks[J]. Frontiers in Marine Science, 2022, 9: 860403. 本文引用 [1]

[23]	李明慧. 基于深度学习的SAR影像海冰分类研究[D]. 上海: 上海海洋大学, 2019. LI M H. Sea ice classification based on deep learning with SAR imagery[D]. Shanghai: Shanghai Ocean University, 2019. 本文引用 [1]

[24]

葛梦滢, 高稳, 祝敏, 等. 基于SE-ConvLSTM的时空特征融合SAR图像海冰分类[J]. 遥感技术与应用, 2023, 38(6):1306-1316.

https://doi.org/10.11873/j.issn.1004-0323.2023.6.1306

摘要

基于合成孔径雷达（SAR）图像的海冰分类已经成为海冰监测的重要基础，但现有方法往往利用图像空间特征，很少考虑时间特征。提出了一种融合时空特征的SAR图像海冰分类网络SE-ConvLSTM。首先使用ConvLSTM对HH和HV极化图像分别提取时空特征，然后将提取的不同层次和通道的时空特征进行拼接，并利用SE通道注意力进行通道特征响应的自适应重新校准，最后利用SoftMax函数进行图像分类。将SI-STSAR-7数据集6个时间步长的图像块作为输入对所提方法与其他分类方法进行了对比实验。结果显示：SE-ConvLSTM在总体情况和分类困难的厚一年冰上分别达到了97.06%和90.01%的精度，表明加入时间信息有助于提高分类准确率。同时，所提网络在生成海冰分布图时对主要冰类型密集度较低的区域和SAR影像的边界位置都具有更好的识别能力。

M Y

, GAO

, ZHU

, et al. Sea ice classification of SAR images based on SE-ConvLSTM spatial-temporal feature fusion[J]. Remote Sensing Technology and Application, 2023, 38(6): 1306-1316.

本文引用 [1]

[25]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA, 2017. 本文引用 [1]

[26]	MU B, LUO X D, YUAN S J, et al. IceTFT v1.0.0: Interpretable long-term prediction of Arctic sea ice extent with deep learning[J]. Geoscientific Model Development, 2023, 16(16): 4677-4697. 本文引用 [1]

[27]	刘剑锋, 郜利康, 赫晓慧, 等. 结合卷积网络与注意力机制的冰凌提取算法[J]. 遥感信息, 2023, 38(4):49-56. LIU J F, GAO L K, HE X H, et al. Combining CNN with self-attention mechanism for ice extraction[J]. Remote Sensing Information, 2023, 38(4): 49-56. 本文引用 [1]

[28]	SUDAKOW I, ASARI V K, LIU R X, et al. MeltPondNet: A swin transformer U-Net for detection of melt ponds on Arctic sea ice[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2022, 15: 8776-8784. 本文引用 [3]

[29]	ZHU Q, GUO H D, ZHANG L, et al. GLA-STDeepLab: SAR enhancing glacier and ice shelf front detection using Swin-TransDeepLab with global-local attention[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 5218113. 本文引用 [2]

[30]	LIU Z, LIN Y T, CAO Y, et al. Swin Transformer: Hierarchical vision Transformer using shifted windows[C]//2021 IEEE/CVF International Conference on Computer Vision (ICCV). Montreal, QC, Canada. IEEE, 2021: 9992-10002. 本文引用 [1]

[31]	SHI W T, XU J, GAO P. SSformer: A lightweight transformer for semantic segmentation[C]// IEEE 24th International Workshop on Multimedia Signal Processing (MMSP). Shanghai, China. IEEE, 2022. 本文引用 [1]

[32]	DAN Y P, ZHU Z N, JIN W S, et al. S-Swin Transformer: Simplified Swin Transformer model for offline handwritten Chinese character recognition[J]. PeerJ Computer Science, 2022, 8: e1093. 本文引用 [1]

[33]	ZHU Y P, LU S. Swin-VoxelMorph: A symmetric unsuper-vised learning model for deformable medical image registration using Swin Transformer[C]// Medical Image Computing and Computer-Assisted Intervention-MICCAI, 2022, 13436:78-87. 本文引用 [1]

[34]	于皓, 田忠翔, 李春花. 基于Sentinel-1双极化数据的北极海域假彩色图像合成方法[J]. 海洋预报, 2022, 39(5):60-69. YU H, TIAN Z X, LI C H. A method of synthesizing RGB pseudo color images based on Sentinel-1 dual-polarization data in the Arctic[J]. Marine Forecasts, 2022, 39(5): 60-69. 本文引用 [1]

[35]	ZHANG J D, ZHANG W Y, HU Y X, et al. An improved sea ice classification algorithm with Gaofen-3 dual-polarization SAR data based on deep convolutional neural networks[J]. Remote Sensing, 2022, 14(4): 906. 本文引用 [1]

[36]	张珂, 冯晓晗, 郭玉荣, 等. 图像分类的深度卷积神经网络模型综述[J]. 中国图象图形学报, 2021, 26(10):2305-2325. ZHANG K, FENG X H, GUO Y R, et al. Overview of deep convolutional neural net-works for image classification[J]. Journal of lmage and Graphics, 2021, 26(10): 2305-2325. 本文引用 [1]

[37]

刘文婷, 卢新明. 基于计算机视觉的Transformer研究进展[J]. 计算机工程与应用, 2022, 58(6):1-16.

https://doi.org/10.3778/j.issn.1002-8331.2106-0442

摘要

Transformer是一种基于自注意力机制、并行化处理数据的深度神经网络。近几年基于Transformer的模型成为计算机视觉任务的重要研究方向。针对目前国内基于Transformer综述性文章的空白，对其在计算机视觉上的应用进行概述。回顾了Transformer的基本原理，重点介绍了其在图像分类、目标检测、图像分割等七个视觉任务上的应用，并对效果显著的模型进行分析。最后对Transformer在计算机视觉中面临的挑战以及未来的发展趋势进行了总结和展望。

LIU

W T

, LU

X M

. Research progress of Transformer based on computer vision[J]. Computer Engineering and Applications, 2022, 58(6): 1-16.

https://doi.org/10.3778/j.issn.1002-8331.2106-0442

本文引用 [1] 摘要

Transformer is a deep neural network based on the self-attention mechanism and parallel processing data. In recent years, Transformer-based models have emerged as an important area of research for computer vision tasks. Aiming at the current blanks in domestic review articles based on Transformer, this paper covers its application in computer vision. This paper reviews the basic principles of the Transformer model, mainly focuses on the application of seven visual tasks such as image classification, object detection and segmentation, and analyzes Transformer-based models with significant effects. Finally, this paper summarizes the challenges and future development trends of the Transformer model in computer vision.