扫 描 看 全 文
彭龙康, 刘励聪, 陈学泓, 等. 遥感影像云检测网络泛化性能研究：以DeepLabv3+为例[J]. 遥感学报, 2021,25(5):1169-1186.
Longkang PENG, Licong LIU, Xuehong CHEN, et al. Generalization ability of cloud detection network for satellite imagery based on DeepLabv3+[J]. National Remote Sensing Bulletin, 2021,25(5):1169-1186.
近年来，深度学习算法得到了长足的发展，并开始应用于云检测。但是深度神经网络模型参数众多，依赖大量训练样本，因此理解其泛化性能对于深度学习在不同遥感影像的实际应用具有重要的参考价值。本文以深度语义分割算法DeepLabv3+为例，以一组广泛使用的云标记数据集“L8 Biome”为验证数据，探讨该算法用于云检测时在不同的地表景观、空间分辨率和光谱波段组合的遥感影像上的泛化性能。云标记数据集“L8 Biome”包含96景具有全球代表性的Landsat 8 OLI影像及相应的人工云掩膜，被广泛用于测试云和云阴影检测算法性能。首先，利用Landsat 8 OLI云标记数据集“L8 Biome”，构建不同类型景观、不同空间分辨率、不同波段组合的训练影像集和测试影像集；其次基于不同训练样本集和测试集，评估了DeepLabv3+算法在不同情况下的云检测精度，并与Fmask算法作对比分析。研究结果表明：（1）使用全混合景观类型的训练集训练出来的云检测网络在总体检测精度（92.81%）与稳定度（标准差12.08%）上都优于使用单一景观类型的训练集训练得到的云检测网络，也优于Fmask的总体精度（88.75%）与稳定度（标准差17.34%），说明在构建深度学习算法的训练集时，应该尽可能包含多类型的地表景观；（2）将全混合景观训练集中剔除一类景观的样本（冰/雪景观除外）构建的“混合-1”训练集与全混合景观训练集训练的DeepLabv3+网络的云检测精度也相差不大，说明现有训练样本集已具备较强的景观泛化能力；（3）基于30 m空间分辨率的全混合景观训练样本集训练得到的DeepLabv3+云检测网络在不同分辨率（30 m、60 m、120 m、240 m）的测试集上云检测精度差异不大，都取得较好的效果，说明DeepLabv3+能够泛化应用于不同空间分辨率的遥感影像，相反Fmask直接应用于低分辨率影像时精度明显下降；（4）DeepLabv3+能充分自适应不同波段的信息用于云检测，总体来说更多的光谱波段输入能够提高DeepLabv3+的云检测的精度和稳定度，其中短波红外波段对于DeepLabv3+区分冰/雪与云具有重要价值，而热红外波段对DeepLabv3+云检测网络的性能提升很微小。以上结果说明利用现有数据集“L8 Biome”训练的DeepLabv3+云检测网络能够适用于多种类型的遥感影像，并优于Fmask算法。
Deep learning algorithms have been developed and applied in detecting clouds for satellite imagery in recent years. However, deep neural network models consist of thousands or millions of parameters, thus usually requiring large amounts of training data. Therefore, understanding the generalization ability of deep learning techniques is vital in their application to cloud detection of different types of satellite imagery. Taking DeepLabv3+, a typical deep semantic segmentation algorithm, as an example, this study explored the generalization ability of the algorithm on the cloud detection of satellite imagery with different landscapes, spatial resolutions, and spectral band combinations based on the cloud labeled dataset “L8 Biome.” The “L8 Biome” dataset consists of 96 typical Landsat 8 OLI images and the corresponding manual cloud mask, which has been widely used for evaluating the performance of cloud and cloud shadow detection methods. First, the cloud labeled dataset “L8 Biome” was used to generate different training and test samples with different landscapes, spatial resolutions, and band combinations. Then, the performance of DeepLabv3+ was evaluated based on different training and test sets and compared with that of the typical Function of Mask (Fmask) algorithm. Results show the following: (1) the DeepLabv3+ trained by a fully mixed training set (consisting of imagery captured over all landscapes) has higher overall cloud accuracy (92.81%) and stability (standard deviation 12.08%) than that trained by the training set of imagery captured over a single landscape and performed better than the Fmask algorithm with an overall cloud accuracy of 88.75% and stability of 17.34%, indicating that the training set of the deep learning algorithm should include images captured over various landscape types; (2) the DeepLabv3+ trained by “mixed-1” training sets, which were built by removing the images captured over a single landscape type (except of snow/ice) from the fully mixed training set, achieved comparable accuracies with that trained by a fully mixed training set, indicating that the available training set is diverse enough and has a good generalization ability on imagery captured over different landscapes; (3) the DeepLabv3+ trained by a fully mixed training set with a 30-m resolution achieved similar cloud detection accuracies on the satellite images with different spatial resolutions (i.e., 30, 60, 120, and 240 m), indicating that the trained DeepLabv3+ could be directly applied on satellite imagery with different spatial resolution, whereas the Fmask algorithm performed poorly on the images with coarse resolutions; (4) DeepLabv3+ could explore the effective information from different spectral bands for cloud detection, and more spectral band input can generally improve the overall cloud accuracy and stability of DeepLabv3+. Among all input bands, shortwave infrared bands are greatly helpful for distinguishing snow/ice from clouds, whereas the thermal infrared band marginally improves the cloud detection accuracy for DeepLabv3+. Results indicate that the DeepLabv3+ cloud detection network trained by the “L8 Biome” dataset can be applied to various types of satellite imagery and outperforms the Fmask algorithm.
深度学习云检测DeepLabv3+泛化性能地表景观光谱波段组合空间分辨率
deep learningcloud detectionDeepLabv3+generalization abilitylandscapespectral band combinationspatial resolution
Ball J E, Anderson D T and Chan C S. 2017. Comprehensive survey of deep learning in remote sensing: theories, tools, and challenges for the community. Journal of Applied Remote Sensing, 11(4): 042609 [DOI: 10.1117/1.JRS.11.042609http://dx.doi.org/10.1117/1.JRS.11.042609]
Chen L C, Papandreou G, Schroff F and Adam H. 2017. Rethinking atrous convolution for semantic image segmentation[EB/OL]. (2017-12-05). http://arxiv.org/pdf/1706.05587v3http://arxiv.org/pdf/1706.05587v3
Chen L C, Zhu Y K, Papandreou G, Schroff F and Adam H. 2018. Encoder-decoder with atrous separable convolution for semantic image segmentation//Proceedings of the 15th European Conference on Computer Vision. Munich, Germany: Springer: 833-851 [DOI: 10.1007/978-3-030-01234-2_49http://dx.doi.org/10.1007/978-3-030-01234-2_49]
Chen Q, Zheng L J, Li X J, Xu C B, Wu Y, Xie D H and Liu L. 2019. Water body extraction from high-resolution satellite remote sensing images based on deep learning. Geography and Geo-information Science, 35(4): 43-49
陈前, 郑利娟, 李小娟, 徐崇斌, 吴俣, 谢东海, 刘亮. 2019. 基于深度学习的高分遥感影像水体提取模型研究. 地理与地理信息科学, 35(4): 43-49[DOI: 10.3969/j.issn.1672-0504.2019.04.007http://dx.doi.org/10.3969/j.issn.1672-0504.2019.04.007]
Chen S L, Chen X H, Chen J, Jia P F, Cao X and Liu C Y. 2016. An iterative haze optimized transformation for automatic cloud/haze detection of Landsat imagery. IEEE Transactions on Geoscience and Remote Sensing, 54(5): 2682-2694 [DOI: 10.1109/TGRS.2015.2504369http://dx.doi.org/10.1109/TGRS.2015.2504369]
Chen X D, Zhang X, Liu L Y and Wang X F. 2019. Enhanced multi-temporal cloud detection algorithm for optical remote sensing images. Journal of Remote Sensing, 23(2): 280-290
陈曦东, 张肖, 刘良云, 汪晓帆. 2019. 增强型多时相云检测. 遥感学报, 23(2): 280-290 [DOI: 10.11834/jrs.20198017http://dx.doi.org/10.11834/jrs.20198017]
Foga S, Scaramuzza P L, Guo S, Zhu Z, Dilley Jr R D, Beckmann T, Schmidt G L, Dwyer J L, Joseph Hughes M and Laue B. 2017. Cloud detection algorithm comparison and validation for operational Landsat data products. Remote Sensing of Environment, 194: 379-390 [DOI: 10.1016/j.rse.2017.03.026http://dx.doi.org/10.1016/j.rse.2017.03.026]
Frantz D, Haß E, Uhl A, Stoffels J and Hill J. 2018. Improvement of the Fmask algorithm for Sentinel-2 images: separating clouds from bright surfaces based on parallax effects. Remote Sensing of Environment, 215: 471-481 [DOI: 10.1016/j.rse.2018.04.046http://dx.doi.org/10.1016/j.rse.2018.04.046]
Harshvardhan, Randall D A and Corsetti T G. 1989. Earth radiation budget and cloudiness simulations with a general circulation model. Journal of the Atmospheric Sciences, 46(13): 1922-1942 [DOI: 10.1175/1520-0469(1989)046<1922:ERBACS>2.0.CO;2http://dx.doi.org/10.1175/1520-0469(1989)046<1922:ERBACS>2.0.CO;2]
Hughes M J and Hayes D J. 2014. Automated detection of cloud and cloud shadow in single-date Landsat imagery using neural networks and spatial post-processing. Remote Sensing, 6(6): 4907-4926 [DOI: 10.3390/rs6064907http://dx.doi.org/10.3390/rs6064907]
Irish R R. 2000. Landsat 7 automatic cloud cover assessment//Proceedings of SPIE 4049, Algorithms for Multispectral, Hyperspectral, and Ultraspectral Imagery VI. Orlando, USA: SPIE: 348-355 [DOI: 10.1117/12.410358http://dx.doi.org/10.1117/12.410358]
Jedlovec G J, Haines S L and LaFontaine F J. 2008. Spatial and temporal varying thresholds for cloud detection in GOES imagery. IEEE Transactions on Geoscience and Remote Sensing, 46(6): 1705-1717 [DOI: 10.1109/TGRS.2008.916208http://dx.doi.org/10.1109/TGRS.2008.916208]
Jeppesen J H, Jacobsen R H, Inceoglu F and Toftegaard T S. 2019. A cloud detection algorithm for satellite imagery based on deep learning. Remote Sensing of Environment, 229: 247-259 [DOI: 10.1016/j.rse.2019.03.039http://dx.doi.org/10.1016/j.rse.2019.03.039]
Lee J, Weger R C, Sengupta S K and Welch R M. 1990. A neural network approach to cloud classification. IEEE Transactions on Geoscience and Remote Sensing, 28(5): 846-855 [DOI: 10.1109/36.58972http://dx.doi.org/10.1109/36.58972]
Li Z W, Shen H F, Cheng Q, Liu Y H, You S C and He Z Y. 2019. Deep learning based cloud detection for medium and high resolution remote sensing images of different sensors. ISPRS Journal of Photogrammetry and Remote Sensing, 150: 197-212 [DOI: 10.1016/j.isprsjprs.2019.02.017http://dx.doi.org/10.1016/j.isprsjprs.2019.02.017]
Li Z W, Shen H F, Li H F, Xia G S, Gamba P and Zhang L P. 2017. Multi-feature combined cloud and cloud shadow detection in GaoFen-1 wide field of view imagery. Remote Sensing of Environment, 191: 342-358 [DOI: 10.1016/j.rse.2017.01.026http://dx.doi.org/10.1016/j.rse.2017.01.026]
Lyu M M, Han L J, Tian S F, Zhou W Q, Li W F and Qian Y G. 2016. Cloud detection under varied surfaces and atmospheric conditions with MODIS imagery. Journal of Remote Sensing, 20(6): 1371-1380
吕明明, 韩立建, 田淑芳, 周伟奇, 李伟峰, 钱雨果. 2016. 多样地表和大气状况下的MODIS数据云检测. 遥感学报, 20(6): 1371-1380 [DOI: 10.11834/jrs.20165281http://dx.doi.org/10.11834/jrs.20165281]
Mateo-García G, Laparra V, López-Puigdollers D and Gómez-Chova L. 2020. Transferring deep learning models for cloud detection between Landsat-8 and Proba-V. ISPRS Journal of Photogrammetry and Remote Sensing, 160: 1-17 [DOI: 10.1016/j.isprsjprs.2019.11.024http://dx.doi.org/10.1016/j.isprsjprs.2019.11.024]
Papadomanolaki M, Vakalopoulou M, Zagoruyko S and Karantzalos K. 2016. Benchmarking deep learning frameworks for the classification of very high resolution satellite multispectral data. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 3(7): 83-88 [DOI: 10.5194/isprs-annals-III-7-83-2016http://dx.doi.org/10.5194/isprs-annals-III-7-83-2016]
Prasad A K, Chai L, Singh R P and Kafatos M. 2006. Crop yield estimation model for Iowa using remote sensing and surface parameters. International Journal of Applied Earth Observation and Geoinformation, 8(1): 26-33 [DOI: 10.1016/j.jag.2005.06.002http://dx.doi.org/10.1016/j.jag.2005.06.002]
Qiu S, Zhu Z and He B B. 2019. Fmask 4.0: improved cloud and cloud shadow detection in Landsats 4-8 and Sentinel-2 imagery. Remote Sensing of Environment, 231: 111205 [DOI: 10.1016/j.rse.2019.05.024http://dx.doi.org/10.1016/j.rse.2019.05.024]
Roy D P, Ju J C, Kline K, Scaramuzza P L, Kovalskyy V, Hansen M, Loveland T R, Vermote E and Zhang C S. 2010. Web-enabled Landsat Data (WELD): landsat ETM+ composited mosaics of the conterminous United States. Remote Sensing of Environment, 114(1): 35-49 [DOI: 10.1016/j.rse.2009.08.011http://dx.doi.org/10.1016/j.rse.2009.08.011]
Saunders R W and Kriebel K T. 1988. An improved method for detecting clear sky and cloudy radiances from AVHRR data. International Journal of Remote Sensing, 9(1): 123-150 [DOI: 10.1080/014 31168808954841http://dx.doi.org/10.1080/01431168808954841]
Schmidhuber J. 2015. Deep learning in neural networks: an overview. Neural Networks, 61: 85-117 [DOI: 10.1016/j.neunet.2014.09.003http://dx.doi.org/10.1016/j.neunet.2014.09.003]
Shan N, Zheng T Y and Wang Z S. 2009. High-speed and high-accuracy algorithm for cloud detection and its application. Journal of Remote Sensing, 13(6): 1138-1155
单娜, 郑天垚, 王贞松. 2009. 快速高准确度云检测算法及其应用. 遥感学报, 13(6): 1138-1155 [DOI: 10.11834/jrs.20090612http://dx.doi.org/10.11834/jrs.20090612]
Shen H F, Li X H, Cheng Q, Zeng C, Yang G, Li H F and Zhang L P. 2015. Missing information reconstruction of remote sensing data: a technical review. IEEE Geoscience and Remote Sensing Magazine, 3(3): 61-85 [DOI: 10.1109/MGRS.2015.2441912http://dx.doi.org/10.1109/MGRS.2015.2441912]
Stowe L L, Davis P A and McClain E P. 1999. Scientific basis and initial evaluation of the CLAVR-1 global clear/cloud classification algorithm for the advanced very high resolution radiometer. Journal of Atmospheric and Oceanic Technology, 16(6): 656-681 [DOI: 10.1175/1520-0426(1999)016<0656:SBAIEO>2.0.CO;2http://dx.doi.org/10.1175/1520-0426(1999)016<0656:SBAIEO>2.0.CO;2]
Sun L, Yang X, Jia S F, Jia C, Wang Q, Liu X Y, Wei J and Zhou X Y. 2020. Satellite data cloud detection using deep learning supported by hyperspectral data. International Journal of Remote Sensing, 41(4): 1349-1371 [DOI: 10.1080/01431161.2019.1667548http://dx.doi.org/10.1080/01431161.2019.1667548]
Verbesselt J, Hyndman R, Newnham G and Culvenor D. 2010. Detecting trend and seasonal changes in satellite image time series. Remote Sensing of Environment 114(1): 106-115 [DOI: 10.1016/j.rse.2009.08.014http://dx.doi.org/10.1016/j.rse.2009.08.014]
Wang J Q, Li J S, Zhou H C and Zhang X. 2019. Typical element extraction method of remote sensing image based on Deeplabv3 + and CRF. Computer Engineering, 45(10): 260-265, 271
王俊强, 李建胜, 周华春, 张旭. 2019. 基于Deeplabv3+与CRF的遥感影像典型要素提取方法. 计算机工程, 45(10): 260-265, 271 [DOI: 10.19678/j.issn.1000-3428.0053359http://dx.doi.org/10.19678/j.issn.1000-3428.0053359]
Yang W, Zhang J L, Xu Z Y and Hu K. 2019. Real-time DeepLabv3+ for pedestrian segmentation. Journal of Optical Technology, 86(9): 570-578 [DOI: 10.1364/JOT.86.000570http://dx.doi.org/10.1364/JOT.86.000570]
Zhang Y C, Rossow W B, Lacis A A, Oinas V and Mishchenko M I. 2004. 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. Journal of Geophysical Research, 109(D19): D19105 [DOI: 10.1029/2003J D004457http://dx.doi.org/10.1029/2003JD004457]
Zhu X L and Helmer E H. 2018a. An automatic method for screening clouds and cloud shadows in optical satellite image time series in cloudy regions. Remote Sensing of Environment, 214: 135-153 [DOI: 10.1016/j.rse.2018.05.024http://dx.doi.org/10.1016/j.rse.2018.05.024]
Zhu Z, Qiu S, He B B and Deng C B. 2018b. Cloud and cloud shadow detection for Landsat images: the fundamental basis for analyzing Landsat time series//Weng Q H, ed. Remote Sensing Time Series Image Processing. Boca Raton: CRC Press: 25-46
Zhu Z, Wang S X and Woodcock C E. 2015. Improvement and expansion of the Fmask algorithm: cloud, cloud shadow, and snow detection for Landsats 4-7, 8, and Sentinel 2 images. Remote Sensing of Environment, 159: 269-277 [DOI: 10.1016/j.rse.2014.12.014http://dx.doi.org/10.1016/j.rse.2014.12.014]
Zhu Z and Woodcock C E. 2012. Object-based cloud and cloud shadow detection in Landsat imagery. Remote Sensing of Environment, 118: 83-94 [DOI: 10.1016/j.rse.2011.10.028http://dx.doi.org/10.1016/j.rse.2011.10.028]
Zi Y, Xie F Y and Jiang Z G. 2018a. A cloud detection method for Landsat 8 images based on PCANet. Remote Sensing, 10(6): 877 [DOI: 10.3390/rs10060877http://dx.doi.org/10.3390/rs10060877]
相关作者
相关机构
京公网安备11010802024621
