基于DenseNet与注意力机制的遥感影像云检测算法

doi:10.6046/zrzyyg.2021128

摘要
图/表
参考文献
相关文章
Metrics

全文: PDF(6594 KB)

HTML
输出: BibTeX | EndNote (RIS)

摘要

遥感影像云检测是遥感影像处理过程中的第一步,针对传统的云检测算法小块薄云检测效果差的问题,该文提出了一种融合注意力机制的密集连接网络遥感影像云检测方法。首先,将自然资源部国土卫星遥感应用中心提供的影像人工勾取云矢量并制作云标签,再将其进行顺序裁剪、色彩抖动、旋转等预处理,以增广样本量; 然后,将预处理过后的遥感影像及其标签一并输入到以DenseNet作为编码器与解码器的神经网络中,编码器与解码器之间加入级联的空洞卷积模块以增大感受野,双注意力机制与全局上下文建模模块以抑制一些无关的细节信息; 最后,经过实验验证表明其精确率可以达到95%以上,交并比可以达到91%以上,较传统云检测算法有较大提高,可以很好地提取小块薄云。

	服务

	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	刘广进
	王光辉
	毕卫华
	刘慧杰
	杨化超

关键词 ：云检测, DenseNet, 双注意力机制, 全局上下文建模模块, 空洞卷积

Abstract：

The cloud detection of remote sensing images is the first step in the process of remote sensing image processing. To address the problem that the traditional cloud detection algorithm has a poor effect on the detection of small and thin clouds, this paper proposes a cloud detection method for densely connected network remote sensing images based on the attention mechanism. First, cloud vectors are manually checked from the images provided by the Land Satellite Remote Sensing Application Center of the Ministry of Natural Resources and cloud labels are made, and the images are preprocessed by sequential clipping, color jitter, rotation, etc. to enlarge the sample size. Then, the pre-processed remote sensing images and their labels are fed into a neural network with DenseNet as the encoder and decoder, and a cascaded atrous convolution module is added between the encoder and decoder to increase the receptive field, and a dual attention mechanism and a global context modeling module are added to suppress some irrelevant detailed information. Finally, the experimental results showed that the accuracy rate could reach 95% and the intersection over union could reach 91%, which are big improvements over the traditional cloud detection algorithm, and this method performs well in extracting small and thin clouds.

Key words： cloud detection DenseNet dual attention mechanism global context modeling module atrous convolution

收稿日期: 2021-04-23 出版日期: 2022-06-20

ZTFLH:

TP751.1

基金资助:国家重点研发计划项目“集成北斗/Galileo/LiDAR/倾斜摄影的智慧城市三维场景重建关键技术研究”(2017YFE0119600)

通讯作者: 王光辉

作者简介: 刘广进(1998-),男,硕士研究生,研究方向为遥感影像云检测。Email: 1538868186@qq.com。

引用本文:

刘广进, 王光辉, 毕卫华, 刘慧杰, 杨化超. 基于DenseNet与注意力机制的遥感影像云检测算法[J]. 自然资源遥感, 2022, 34(2): 88-96.
LIU Guangjin, WANG Guanghui, BI Weihua, LIU Huijie, YANG Huachao. Cloud detection algorithm of remote sensing image based on DenseNet and attention mechanism. Remote Sensing for Natural Resources, 2022, 34(2): 88-96.

链接本文:

https://www.gtzyyg.com/CN/10.6046/zrzyyg.2021128 或 https://www.gtzyyg.com/CN/Y2022/V34/I2/88

Fig.1 ResNet-Block结构

Fig.2 密集连接模块

Fig.3 转换模块

Fig.4 通道注意力模块

Fig.5 位置注意力模块

Fig.6 非局部神经网络

Fig.7 全局上下文建模模块

Fig.8 空洞卷积模块

Fig.9 融入注意力机制的密集连接网络

Tab.1 原始影像及其真值标签

Tab.2 增强之后的影像及真值标签

Tab.3 训练集上的不同算法模型评价结果

Tab.4 验证集上的不同算法模型评价结果

Fig.10 AD-DenseNet算法损失值与交并比随epoch的变化曲线

Fig.11 AD-DenseNet算法精确率与召回率随epoch的变化曲线

Fig.12 不同算法的云检测结果对比

[1]	Zhang Y, Rossow W B, Lacis A, et al. Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets:Refinements of the radiative transfer model and the input data[J]. Journal of Geophysical Research:Atmospheres, 2004, 109(d19):105-115.
[2]	Ju J C, Roy D P. The availability of cloud-free Landsat ETM+ data over the conterminous United States and globally[J]. Remote Sensing of Environment, 2007, 112(3):1196-1211. doi: 10.1016/j.rse.2007.08.011
[3]	Xie F, Shi M Y, Shi Z, 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, 10:3631-3640. doi: 10.1109/JSTARS.2017.2686488
[4]	Zhu Z, Woodcock C E. Object-based cloud and cloud shadow detection in Landsat imagery[J]. Remote Sensing of Environment, 2012, 118:83-94. doi: 10.1016/j.rse.2011.10.028
[5]	Otsu N. A Threshold selection method from gray-level histograms[J]. IEEE Transactions on Systems Man and Cybernetics, 2007, 9(1):62-66. doi: 10.1109/TSMC.1979.4310076
[6]	李玉功. K-means聚类算法的改进及其在彩色图像分割中的应用[D]. 合肥: 安徽大学, 2018.
	Li Y G. Improvement of K-means clustering algorithm and its application in color image segmentation[D]. Hefei: Anhui University, 2018.
[7]	康一飞, 潘励, 孙明伟, 等. 基于高斯混合模型法的国产高分辨率卫星影像云检测[J]. 武汉大学学报(信息科学版), 2017, 42(6):782-788.
	Kang Y F, Pan L, Sun M W, et al. Gaussian mixture model based cloud detection for Chinese high resolution satellite imagery[J]. Geomatics and Information Science of Wuhan University, 2017, 42(6):782-788.
[8]	仇一帆, 柴登峰. 无人工标注数据的Landsat影像云检测深度学习方法[J]. 国土资源遥感, 2021, 33(1):102-107.doi: 10.6046/gtzyyg.2020090. doi: 10.6046/gtzyyg.2020090
	Qiu Y F, Chai D F. A deep learning method for Landsat image cloud detection without manually labeled data[J]. Remote Sensing for Land and Resources, 2021, 33(1):102-107.doi: 10.6046/gtzyyg.2020090. doi: 10.6046/gtzyyg.2020090
[9]	栗旭升, 刘玉峰, 陈冬花, 等. 结合图像特征的支持向量机高分一号云检测[J]. 国土资源遥感, 2020, 32(3):55-62.doi: 10.6046/gtzyyg.2020.03.08. doi: 10.6046/gtzyyg.2020.03.08
	Li X S, Liu Y F, Chen D H, et al. Cloud detection based on support vector machine with image features for GF-1 data[J]. Remote Sensing for Land and Resources, 2020, 32(3):55-62.doi: 10.6046/gtzyyg.2020.03.08. doi: 10.6046/gtzyyg.2020.03.08
[10]	徐启恒, 黄滢冰, 陈洋. 结合超像素和卷积神经网络的国产高分辨率遥感影像云检测方法[J]. 测绘通报, 2019(1):50-55.
	Xu Q H, Huang Y B, Chen Y. Cloud detection for Chinese high resolution remote sensing imagery using combining superpixel with convolution neural network[J]. Bulletin of Surveying and Mapping, 2019(1):50-55.
[11]	刘云峰, 杨珍, 韩骁, 等. 国产高分辨率遥感卫星影像云检测方法分析[J]. 测绘通报, 2020(11):66-70.
	Liu Y F, Yang Z, Han X, et al. Research on cloud detection method of domestic high-resolution satellite images[J]. Bulletin of Surveying and Mapping, 2020(11):66-70.
[12]	张家强, 李潇雁, 李丽圆, 等. 基于深度残差全卷积网络的Landsat8遥感影像云检测方法[J]. 激光与光电子学进展, 2020, 57(10):364-371.
	Zhang J Q, Li X Y, Li L Y, et al. Landsat8 remote sensing image based on deep residual fully convolutional network[J]. Laser and Optoelectronics Progress, 2020, 57(10):364-371.
[13]	张永宏, 蔡朋雁, 陶润喆, 等. 基于改进U-Net网络的遥感图像云检测[J]. 测绘通报, 2020(3):17-20.
	Zhang Y H, Cai P Y, Tao R Z, et al. Cloud detection for remote sensing images using improved U-Net[J]. Bulletin of Surveying and Mapping, 2020(3):17-20.
[14]	康超萌. 基于神经网络的国产高分光学遥感图像云检测[D]. 西安: 中国科学院大学(中国科学院西安光学精密机械研究所), 2018:20-27.
	Kang C M. Cloud detection in domestic high-resolution remote sensing image based deep neural networks[D]. Xi’an: University of Chinese Academy of Sciences, 2018:20-27.
[15]	张晨, 张秀再, 杨昌军. 双注意力RDA-Net遥感影像云与云阴影检测方法[J/OL]. 激光与光电子学进展, 2021:1-19. https://kns.cnki.net/kcms/detail/31.1690.TN.20210222.1114.023.html.
	Zhang C, Zhang X Z, Yang C J. Dual attention RDA-Net for cloud and cloud shadow detection of remote sensing image[J/OL]. Laser and Optoelectronics Progress, 2021:1-19. https://kns.cnki.net/kcms/detail/31.1690.TN.20210222.1114.023.html.
[16]	Zhou L C, Zhang C, Wu M. D-LinkNet:LinkNet with pretrained encoder and dilated convolution for high resolution satellite imagery road extraction[C]// IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2018:182-186.
[17]	彭博. 基于深度学习的遥感图像道路信息提取算法研究[D]. 成都: 电子科技大学, 2019:45-46.
	Peng B. Research on road information extraction of remote sensing image based on deep learning[D]. Chengdu: University of Electronic Science and Technology of China, 2019:45-46.
[18]	Cao Y, Xu J R, Stephen L, et al. GCNet:Non-local networks meet squeeze-excitation networks and beyond[C]// IEEE CVF International Conference on Computer Vision (ICCV), 2019:1971-1980.
[19]	Fu J, Liu J, Tian H, et al. Dual attention network for scene segmentation[C]// IEEE Conference on Computer Vision and Pattern Recognition, 2019:3146-3154.
[20]	Huang G, Liu Z, Van Der Maaten L, et al. Densely connected convo-lutional networks[C]// Proceedings of the 30th IEEE Conference on Computer Vision and Pattern Recognition(CVPR),Honolulu, 2017.
[21]	He K M, Zhan X Y, Ren S Q. Deep residual learning for image reco-gnition[C]// IEEE Conference on Computer Vision and Pattern Recognition, 2016:770-778.
[22]	Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]// IEEE Conference on Computer Vision and Pattern Recognition(CVPR), 2018:7132-7141.
[23]	Woo S, Park J, Lee J Y, et al. CBAM:Convolutional block attention module[C]// 15th European Conference on Computer Vision (ECCV), 2018, 11211:3-19.
[24]	Wang X L, Girshick R, Gupta A, et al. Non-local neural networks[C]// IEEE Conference on Computer Vision and Pattern Recognition, 2018:7794-7803.
[25]	Chen L C, Papandreou G, Kokkinos I, et al. DeepLab:Semantic image segmentation with deep convolutional nets,atrous convolution,and fully connected CRFs[J]. IEEE Transactions on Pattern Analysis and Machine Interlligence, 2018, 40(4):834-848.

[1]	伍炜超, 叶发旺. 面向多背景环境的Sentinel-2云检测[J]. 自然资源遥感, 2023, 35(3): 124-133.
[2]	鄢俊洁, 郭雪星, 瞿建华, 韩旻. 基于朴素贝叶斯方法的FY-4A/AGRI云检测模型[J]. 自然资源遥感, 2022, 34(3): 33-42.
[3]	栗旭升, 刘玉锋, 陈冬花, 刘赛赛, 李虎. 结合图像特征的支持向量机高分一号云检测[J]. 国土资源遥感, 2020, 32(3): 55-62.
[4]	林志玮, 涂伟豪, 黄嘉航, 丁启禄, 周铮雯, 刘金福. 基于FC-DenseNet的低空航拍光学图像树种识别[J]. 国土资源遥感, 2019, 31(3): 225-233.
[5]	刘紫涵, 吴艳兰. 遥感图像云检测方法研究进展[J]. 国土资源遥感, 2017, 29(4): 6-7.
[6]	夏浪, 毛克彪, 孙知文, 马莹, 赵芬. 基于DNB验证的VIIRS夜间云检测方法[J]. 国土资源遥感, 2014, 26(3): 74-79.
[7]	秦雁, 邓孺孺, 何颖清, 陈蕾, 陈启东, 熊首萍. 基于光谱及几何信息的TM图像厚云去除算法[J]. 国土资源遥感, 2012, 24(4): 55-61.
[8]	韩杰, 杨磊库, 李慧芳, 梁洪有, 马晓红, 谢玉娟. 基于动态阈值的HJ-1B图像云检测算法研究[J]. 国土资源遥感, 2012, 24(2): 12-18.
[9]	侯岳, 刘培洵, 陈顺云, 刘少峰. 基于MODIS影像的夜间云检测算法研究[J]. 国土资源遥感, 2008, 20(1): 34-37.
[10]	何全军, 曹静, 黄江, 吴志军. 基于多光谱综合的MODIS数据云检测研究[J]. 国土资源遥感, 2006, 18(3): 19-22.
[11]	蒋耿明, 牛铮, 阮伟利, 王长耀. MODIS影像合成算法研究和实现[J]. 国土资源遥感, 2004, 16(2): 11-15，79.
[12]	杨俊泉, 沈建中, 莫伟华, 张远飞, 陈尚文, 林少雄. NOAA卫星图像云点检测的三维统计图集合判别[J]. 国土资源遥感, 1997, 9(4): 54-55.

Viewed

Full text

Abstract

Cited

Shared

Discussed