基于深度学习的多角度遥感影像云检测方法

大气与环境光学学报 ›› 2020, Vol. 15 ›› Issue (5): 380-392.

基于深度学习的多角度遥感影像云检测方法

李佳欣1;2, 赵鹏1;2, 方薇3∗, 宋尚香1;2

1 安徽大学计算智能与信号处理教育部重点实验室, 安徽合肥 230039; 2 安徽大学计算机科学与技术学院, 安徽合肥 230001; 3 中国科学院合肥物质科学研究院安徽光学精密机械研究所通用光学定标与表征技术重点实验室, 安徽合肥 230031

收稿日期:2020-02-18 修回日期:2020-03-30 出版日期:2020-09-28 发布日期:2020-09-28
通讯作者: 方薇：E-mail: fwei@aiofm.ac.cn E-mail:fwei@aiofm.ac.cn
作者简介:李佳欣 ( 1994 - ), 男, 湖北天门人, 硕士研究生, 主要从事计算机视觉与遥感影像方面的研究。 E-mail: arthurpendgradon@163.com
基金资助:
The National Natural Science Foundation of China (国家自然科学基金, 61602004), Natural Science Foundation of Anhui Province (安徽省自然科学基金项目, 1908085MF188, 1908085MF182), Natural Science Foundation of the Education Department of Anhui Province (安徽省高校自然科学研究重点项目, KJ2018A0013, KJ2017A011)

Cloud Detection of Multi-Angle Remote Sensing Image Based on Deep Learning

Li Jiaxin1;2, Zhao Peng1;2, Fang Wei3∗, Song Shangxiang1;2 #br#

1 Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education, Anhui University, Hefei 230039, China; 2 School of Computer Science and Technology, Anhui University, Hefei 230001, China; 3 Key Laboratory of Optical Calibration and Characterization of Chinese Academy of Science, Anhui Institute of Optics and Fine Mechanics, HFIPS, Chinese Academy of Sciences, Hefei 230031, China

Received:2020-02-18 Revised:2020-03-30 Published:2020-09-28 Online:2020-09-28
Supported by:
The National Natural Science Foundation of China

摘要/Abstract

摘要： 云检测是遥感影像处理的重要任务之一。目前遥感影像云检测中多使用到卫星的多光谱、多通道信息, 而关于多角度信息对云检测影响的研究较少。为了探索遥感影像多角度信息作为云特征对训练云分类网络精度的影响, 提出一种基于深度学习的遥感多角度云检测方法, 以 SegNet 为基础网络结构, 提取含有多角度信息的遥感影像的特征表示, 训练含有多角度信息的遥感影像云检测模型。测试结果表明, 所提方法全局精度为 91.39%, 平均重叠率为 83.99%。分析表明单角度云检测具有一定的局限性, 而利用多角度信息作为云特征训练云分类网络可以提升云检测精度。此外, 还探索了 POLDER 仪器中不同角度对于云检测结果的影响情况。

关键词: 云检测, 遥感影像, 多角度, 神经网络, SegNet

Abstract: Cloud detection is one of the important tasks for remote sensing image processing. At present, the multi-spectral and multi-channel information is often used in cloud detection of remote sensing image, but the research on the influence of multi-angle information on cloud detection is still insufficient. To explore the influence of multi-angle information as cloud feature on the accuracy of cloud classification, a cloud detection method with multi-angles remote sensing based on deep learning is proposed. The proposed method takes SegNet as backbone network, and trains a multi-angle information based cloud detection model by extracting the remote sensing image feature with multi-angle information. Extensive experimental results demonstrate that the Global Accuracy and the mean intersection over union (MeanIoU) of the proposed method are 91.39% and 83.99% respectively. And the method proves the limitations of single angle cloud detection and the effectiveness of multi-angle information on the improvement of the cloud detection accuracy. In addtion, the influence of different angles on the cloud detection in POLDER is also explored.

Key words: cloud detection, remote sensing image, multi-angle, neural network, SegNet

中图分类号:

TP79

李佳欣, 赵鹏, 方薇∗, 宋尚香, . 基于深度学习的多角度遥感影像云检测方法[J]. 大气与环境光学学报, 2020, 15(5): 380-392.

Li Jiaxin, Zhao Peng, Fang Wei∗, Song Shangxiang, . Cloud Detection of Multi-Angle Remote Sensing Image Based on Deep Learning[J]. Journal of Atmospheric and Environmental Optics, 2020, 15(5): 380-392.

参考文献 J

[1]	Hou Shuwei, Sun Wenfang, Zheng Xiaosong. Overview of remote sensing image cloud detection methods [J]. Space Electronic
	Technology, 2014, 11 (3): 68-76 (in Chinese).
	侯舒维, 孙文方, 郑小松. 遥感图像云检测方法综述 [J]. 空间电子技术, 2014, 11 (3): 68-76.
[2]	Ackerman S A, Strabala K I, Menzel W P, et al. Discriminating clear sky from clouds with MODIS [J]. Journal of Geophysical
	Research, 1998, 103: 32141-32157.
[3]	Jedlovec G. Automated Detection of Clouds in Satellite Imagery [M]. Advances in Geoscience and Remote Sensing. InTech
	Open Access Publisher, 2009.
[4]	Kriebel K T, Gesell G, Kastner M, et al. The cloud analysis tool APOLLO: Improvements and validations [J]. International
	Journal of Remote Sensing, 2003, 24 (12): 2389-2408.
[5]	Buriez J C, Vanbauce C, Parol F, et al. Cloud detection and derivation of cloud properties from POLDER [J]. International
	Journal of Remote Sensing, 1997, 18 (13): 2785-2813.
[6]	Fang Wei, Qiao Yanli, Zhang Dongying, et al. Threshold optimization in cloud detection by polarized multichannel remote
	sensing [J]. Acta Optica Sinica, 2018, 38(12): 1228005 (in Chinese).方薇, 乔延利, 张冬英, 等. 偏振多通道遥感云检测的阈值优化 [J]. 光学学报, 2018, 38 (12): 1228005.
[7]	Xia Yu, Cui Shengcheng, Yang Shizhi. Cloud detection method for high resolution satellite image based on multi-dimensional
	features [J]. Journal of Atmospheric and Environmental Optics, 2017, 12 (06): 69-77 (in Chinese).
	夏雨, 崔生成, 杨世植. 综合高分卫星图像多维特征的云检测方法 [J]. 大气与环境光学学报, 2017, 12(06) : 69-77.
[8]	Xie F Y, Shi M Y, Shi Z W, et al. Multilevel cloud detection in remote sensing images based on deep learning [J]. IEEE Journal
	of Selected Topics in Applied Earth Observations and Remote Sensing, 2017, 10 (8): 3631-3640.
[9]	Zhan Y J, Wang J, Shi J P, et al. Distinguishing cloud and snow in satellite images via deep convolutional network [J]. IEEE
	Geoscience & Remote Sensing Letters, 2017, 14 (10): 1785-1789.
[10]	Chen Yang, Fan Rongshuang, Wang Jingxue, et al. Deep learning based cloud detection for ZY-3 remote sensing image [J].
	Acta Optica Sinica, 2018, 38 (01): 0128005 (in Chinese).
	陈洋, 范荣双, 王竞雪, 等. 基于深度学习的资源三号卫星遥感影像云检测方法 [J]. 光学学报, 2018, 38 (01): 0128005.
[11]	Gao Xianjun, Wan Youchuan, Zheng Shunyi, et al. Real-time automatic cloud detection during the process of taking aerial
	photographs [J]. Spectroscopy and Spectral Analysis, 2014, 34 (7): 1909-1913 (in Chinese).
	高贤君, 万幼川, 郑顺义, 等. 航空摄影过程中云的实时自动检测 [J]. 光谱学与光谱分析, 2014, 34 (7): 1909-1913.
[12]	Sun Xiaobing, Hong Jin, Qiao Yanli, et al. Project of multi-angle polarimetric system onboard satellite for remote sensing of
	atmosphere [J]. Journal of Atmospheric and Environmental Optics, 2006, 1 (3): 198-201 (in Chinese).
	孙晓兵, 洪津, 乔延利, 等. 卫星大气多角度偏振遥感系统方案研究 [J]. 大气与环境光学学报, 2006, 1 (3): 198-201.
[13]	Buriez J C, Vanbauce C, Parol F, et al. Cloud detection and derivation of cloud properties from POLDER [J]. International
	Journal of Remote Sensing, 1997, 18 (13): 2785-2813.
[14]	Kang Qing, Yuan Yinlin, Weng Jianwen, et al. System-level polarized calibration methods in laboratory of directional polarization camera [J]. Journal of Atmospheric and Environmental Optics. 2019, 14 (1): 36-48 (in Chinese).
	康晴, 袁银麟, 翁建文, 等. 多角度偏振成像仪系统级偏振定标方法研究 [J]. 大气与环境光学学报. 2019, 14 (1): 36-48.
[15]	Zheng Fengxun, Hou Weizhen, Li Zhengqiang. Optimal estimation retrieval for directional polarimetric camera onboard
	Chinese Gaofen-5 satellite: an analysis on multi-angle dependence and a posterior error [J]. Acta Physica Sinica. 2019, 68
(4)	: 040701 (in Chinese).
	郑逢勋, 侯伟真, 李正强. 高分五号卫星多角度偏振相机最优化估计反演:角度依赖和后验误差分析 [J]. 物理学报, 2019,
68	(4): 040701.
[16]	Chen Hongbin, Fan Xuehua, Han Zhigang. Progress of remote sensing observation and measurement with POLDER multiangle and multi-channel partial vibration detector [J]. Remote Sensing Technology and Application, 2006, 21 (2): 83-92 (in
	Chinese).
	陈洪滨, 范学花, 韩志刚. POLDER 多角度多通道偏振探测器对地遥感观测研究进展 [J]. 遥感技术与应用, 2006, 21
(2)	: 83-92.
[17]	Badrinarayanan V, Kendall A, Cipolla R. SegNet: a deep convolutional encoder-decoder architecture for scene segmentation
[J]	IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39 (12): 2481-2495.