Deep learning classification method of Landsat 8 OLI images based on inaccurate prior knowledge

doi:10.11772/j.issn.1001-9081.2020040446

Abstract

Abstract: Remote sensing image interpretation plays an important role in the acquisition of Land Use and Land Cover (LULC) information, and automatic classification serves as the key to improve the efficiency of LULC information acquisition. The actual scenes have a great mount of inaccurate prior knowledge. Extracting and integrating the available knowledge in the prior knowledge can help to further improve the accuracy, automation rate and scale application ability of image classification methods. Based on the above situation, a new deep learning classification method of Landsat 8 OLI images based on inaccurate prior knowledge was proposed. For the proposed method, inaccurate units in prior knowledge were avoided automatically, realizing automatic region selection and feature extraction of classified samples and obtaining high confidence knowledge in the constraint space of patches. Then, the deep residual network was trained by using these classified samples, and the accurate classification of large-area images was achieved. In the experiment, Xinbei district of Changzhou city was taken as the example, the data of 2009 land use status of this district was selected as the prior data, and the 2014 Landsat 8 OLI image of this district was selected as the to-be-classified image. The experimental results show that the proposed method has advantages such as the integration of inaccurate prior knowledge and the accurate classification of large-area contiguous LULC information. Besides, it can obtain the accurate boundary of main land use patches, and has the accuracy for patch classification in the whole image of 88.7% and the Kappa coefficient of 0.842.The proposed method can cooperate with deep learning method to achieve high precision Landsat 8 OLI remote sensing image classification.

Key words: Land Use and Land Cover (LULC) classification, Landsat 8 OLI, inaccurate prior knowledge, automatic sample selection, sample feature, deep learning

摘要： 遥感影像解译是获得土地利用和土地覆盖（LULC）信息最为重要的途径之一，而自动化分类是提高LULC信息获取效率的关键。实际场景中包含大量不精准的先验知识，提取并融合其中的可用知识能进一步提高影像分类方法的精度、自动化率和规模应用能力。基于上述情况，提出了一种融合不精准先验知识的Landsat 8 OLI影像深度学习分类方法。该方法可自动规避先验知识中的不精准单元，在图斑约束空间内实现了分类样本的自动化区域选择和特征提取，并获得了高置信度知识，然后利用这些分类样本训练深度残差网络，从而实现大区域影像的精确分类。以常州市新北区为例进行实验，选用该区域2009年土地利用现状数据作为先验数据，2014年Landsat 8 OLI影像作为待分类影像。实验结果表明，所提方法可融合不精准先验知识，对大面积连片LULC信息分类精确，主要地类图斑界限准确，全图分类图斑精度达到了88.7%，Kappa系数为0.842。该方法能配合深度学习方法实现高精度Landsat 8 OLI遥感影像分类。

关键词: 土地利用和土地覆盖分类, Landsat 8 OLI, 不精准先验知识, 样本自动选取, 样本特征, 深度学习

CLC Number:

XU Changqing, CHEN Zhenjie, HOU Renfu. Deep learning classification method of Landsat 8 OLI images based on inaccurate prior knowledge[J]. Journal of Computer Applications, 2020, 40(12): 3550-3557.

许长青, 陈振杰, 侯仁福. 融合不精准先验知识的Landsat 8 OLI影像深度学习分类方法[J]. 计算机应用, 2020, 40(12): 3550-3557.

References

[1] SULLA-MENASHE D,GRAY J M,ABERCROMBIE S P,et al. Hierarchical mapping of annual global land cover 2001 to present:the MODIS Collection 6 Land Cover product[J]. Remote Sensing of Environment,2019,222:183-194.
[2] 胡光印, 金会军, 董治宝, 等. 黄河源区土地利用/覆盖变化(LUCC)研究[J]. 冰川冻土, 2014, 36(3):573-581.(HU G Y, JIN H J,DONG Z B,et al. Research of Land-Use and Land-Cover Change(LUCC) in the source regions of the Yellow River[J]. Journal of Glaciology and Geocryology,2014,36(3):573-581.)
[3] SHARMA A,LIU X,YANG X,et al. A patch-based convolutional neuralnetwork for remote sensing image classification[J]. Neural Networks,2017,95:19-28.
[4] SHIMABUKURO Y E,BATISTA G T,MELLO E M K,et al. Using shade fraction image segmentation to evaluate deforestation in Landsat Thematic Mapper images of the Amazon region[J]. International Journal of Remote Sensing,1998,19(3):535-541.
[5] KAARNA A,ZEMCIK P,KALVIAINEN H,et al. Compression of multispectral remote sensing images using clustering and spectral reduction[J]. IEEE Transactions on Geoscience and Remote Sensing,2000,38(2):1073-1082.
[6] PAL M,MATHER P M. An assessment of the effectiveness of decision tree methods for land cover classification[J]. Remote sensing of Environment,2003,86(4):554-565.
[7] OTUKEI J R,BLASCHKE T. Land cover change assessment using decision trees,support vector machines and maximum likelihood classification algorithms[J]. International Journal of Applied Earth Observation and Geoinformation,2010,12(S1):S27-S31.
[8] RODRIGUEZ-GALIANO V F,GHIMIRE B,ROGAN J,et al. An assessment of the effectiveness of a random forest classifier for landcover classification[J]. ISPRS Journal of Photogrammetry and Remote Sensing,2012,67:93-104.
[9] ANTHONY G,GREGG H,TSHILIDZI M. Image classification using SVMs:one-against-one vs one-against-all[EB/OL].[2019-11-19]. https://arxiv.org/pdf/0711.2914.pdf.
[10] KAVZOGLU T, MATHER P. The use of backpropagating artificial neuralnetworks in land cover classification[J]. International Journal of Remote Sensing,2003,24(23):4907-4938.
[11] CASTELLUCCIO M,POGGI G,SANSONE C,et al. Land use classification in remote sensing images by convolutional neuralnetworks[EB/OL].[2019-08-01]. https://arxiv.org/pdf/1508.00092.pdf.
[12] ZHANG C,SARGENT I,PAN X,et al. Joint deep learning for land cover and land use classification[J]. Remote Sensing of Environment,2019,221:173-187.
[13] ZHANG C, SARGENT I, PAN X, et al. An Object-based Convolutional Neural Network (OCNN) for urban land use classification[J]. Remote Sensing of Environment,2018,216:57-70.
[14] ZHANG C,PAN X,LI H,et al. A hybrid MLP-CNN classifier for very fine resolution remotely sensed image classification[J]. ISPRS Journal of Photogrammetry and Remote Sensing,2018, 140:133-144.
[15] 徐小明, 杜自强, 张红, 等. 晋北地区1986~2010年土地利用/覆被变化的驱动力[J]. 中国环境科学, 2016, 36(7):2154-2161. (XU X M,DU Z Q,ZHANG H,et al. Quantitative analysis on driving forces of land use/cover change in north Shanxi province during 1986~2010[J]. China Environmental Science,2016,36(7):2154-2161.)
[16] HOMER C,DEWITZ J,YANG L,et al. Completion of the 2011 national land cover database for the conterminous united states-representing a decade of land cover change information[J]. Photogrammetric Engineering and Remote Sensing,2015,81(5):345-354.
[17] YANG L, MANSARAY L R, HUANG J, et al. Optimal segmentation scale parameter,feature subset and classification algorithm for geographic object-based crop recognition using multisource satellite imagery[J]. Remote Sensing,2019,11(5):Article No. 514.
[18] DOERSCH C,GUPTA A,EFROS A A. Unsupervised visual representation learning by context prediction[C]//Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway:IEEE,2015:1422-1430.
[19] LEE H, HWANG S J, SHIN J. Self-supervised label augmentation via input transformations[EB/OL].[2020-04-10]. https://arxiv.org/pdf/1910.05872.pdf.
[20] ABDI H, WILLIAMS L J. Tukey's Honestly Significant Difference (HSD) test[EB/OL].[2020-04-10]. http://www.docin.com/p-782119702.html.
[21] 周志华. 机器学习[M]. 北京:清华大学出版社, 2016:23-28. (ZHOU Z H. Machine Learning[M]. Beijing:Tsinghua University Press,2016:23-28.)
[22] 李胜. 联合领域知识和深度学习的城市地表覆盖变化检测方法[D]. 武汉:武汉大学, 2018:3-10.(LI S. Detection method of urban land cover change based oncombining domain knowledge and deep learning[D]. Wuhan:Wuhan University,2018:3-10.)
[23] 孙家波. 基于知识的高分辨率遥感影像耕地自动提取技术研究[D]. 北京:中国农业大学, 2014:4-6.(SUN J B. Study on automatic cultivated land extraction from high resolution satellite imagery based on knowledge[D]. Beijing:China Agricultural University,2014:4-6.)
[24] 梁杰, 陈嘉豪, 张雪芹, 等. 基于独热编码和卷积神经网络的异常检测[J]. 清华大学学报(自然科学版), 2019, 59(7):523-529. (LANG J,CHEN J H,ZHANG X Q,et al. One-hot encoding and convolutional neuralnetwork based anomaly detection[J]. Journal of Tsinghua University (Science and Technology),2019,59(7):523-529.)
[25] 李雪菲. 大规模结构化数据特征抽取系统的设计与实现[D]. 南京:南京大学,2019:10-11. (LI X F. Design and implementation of feature extraction system of large-scale structured data[D]. Nanjing:Nanjing University,2019:10-11.)
[26] CHEN T, GUESTRIN C. XGboost:a scalable tree boosting system[C]//Proceedings of the 201622nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York:ACM,2016:785-794.
[27] TOWNSHEND J R G,JUSTICE C O. Analysis of the dynamics of African vegetation using the normalized difference vegetation index[J]. International Journal of Remote Sensing,1986,7(11):1435-1445.
[28] HUETE A R. A Soil-Adjusted Vegetation Index (SAVI)[J].Remote Sensing of Environment,1988,25(3):295-309.
[29] JIANG Z,HUETE A R,DIDAN K,et al. Development of a twoband enhanced vegetation index without a blue band[J]. Remote Sensing of Environment,2008,112(10):3833-3845.
[30] KAUFMAN Y J, TANRE D. Atmospherically Resistant Vegetation Index (ARVI) for EOS-MODIS[J]. IEEE Transactions on Geoscience and Remote Sensing,1992,30(2):261-270.
[31] RIGGS G A,HALL D K,SALOMONSON V V. A snow index for the Landsat Thematic Mapper and moderate resolution imaging spectroradiometer[C]//Proceedings of the 1994 IEEE International Geoscience and Remote Sensing Symposium. Piscataway:IEEE,1994:1942-1944.
[32] XU H. Modification of Normalised Difference Water Index (NDWI) to enhance open water features in remotely sensed imagery[J]. International Journal of Remote Sensing,2006,27(14):3025-3033.
[33] LÓPEZ GARCÍA M J,CASELLES V. Mapping burns and natural reforestation using Thematic Mapper data[J]. Geocarto International,1991,6(1):31-37.
[34] LUTES D C,KEANE R E,CARATTI J F,et al. FIREMON:fire effects monitoring and inventory system[R]. USDA Forest Service Fort Collins:Rocky Mountain Research Station of USDA Forest Service,2006.
[35] SEGAL D. Theoretical basis for differentiation of ferric-iron bearing minerals using Landsat MSS data[C]//Proceedings of the 1982 Symposium for Remote Sensing of Environment/2nd Thematic Conference on Remote Sensing for Exploratory Geology. Ann Arbor:Environmental Research Institute of Michigan, 1982:949-951.
[36] ALASTA A F. Using remote sensing data to identify iron deposits in central western Libya[C]//Proceedings of the 2011 International Conference on Emerging Trends in Computer and Image Processing. Berlin:Springer,2011:56-61.
[37] HE K,ZHANG X,REN S,et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE, 2016:770-778.
[38] IOFFE S,SZEGEDY C. Batch normalization:accelerating deepnetwork training by reducing internal covariate shift[C]//Proceedings of the 201532nd International Conference on International Conference on Machine Learning. New York:JMLR.org,2015:448-456.
[39] ZHANG X,ZHU K,CHEN G,et al. Geospatial object detection on high resolution remote sensing imagery based on double multiscale feature pyramidnetwork[J]. Remote Sensing,2019,11(7):Article No. 755.
[40] WANG G,LI W,AERTSEN M,et al. Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neuralnetworks[J]. Neurocomputing,2019,338:34-45.
[41] WONG R Y,HALL E L. Scene matching with invariant moments[J]. Computer Graphics and Image Processing,1978,8(1):16-24.
[42] 刘晓丽. 基于机器视觉的异常行为检测[D]. 鞍山:辽宁科技大学, 2012:34-38.(LIU X L. Detection of human action based on machine vision[D]. Anshan:University of Science and Technology Liaoning,2012:34-38.)
[43] HUANG B,ZHAO B,SONG Y. Urban land-use mapping using a deep convolutional neuralnetwork with high spatial resolution multispectral remote sensing imagery[J]. Remote Sensing of Environment,2018,214:73-86.