一种利用Landsat年度时序数据的土地覆盖分类方法
|
摘要
针对大面积土地覆盖遥感分类中数据获取难度大、复杂度高、分类结果不够精确且易受季候变化影响等问题,提出了一种利用Landsat时间序列数据,生成年度时序特征,并结合特定算法(UniBagging)进行土地覆盖分类的方法(LandUTime)。该方法定义了一种基于时间序列数据的特征生成方式,根据时序数据特点,设计了一种基于特征子空间的集成分类算法。实现过程分为2个阶段,首先基于特定模型,在像元级别上对Landsat时间序列图像进行回归分析,生成模式特征,然后将所有特征整合成"特征块",根据特征子空间将基分类器集成到相互分离的集合中,最后通过加权投票的方法进行分类结果输出。实验结果与定量分析表明,与传统的特征提取及分类方法相比,该方法提高了分类精度,而且对高维数据具有鲁棒性;可以有效克服大面积土地覆盖分类中云遮掩、数据条带和物候变化等问题的影响,具有较高的准确性和实用性。
To overcome problems,such as difficulty of data acquisition,complex data structure,low classification accuracy and influence of seasonal variation,existing in large area land cover classification,this paper proposes a novel land cover classification method(UniBagging)which takes annual time series Landsat data as input.This method defines a feature generation strategy for time series data,and designes an ensemble classification algorithm based on subspace,according to the characteristics of time series data.The method contains two stages.Firstly,based on specific model,regression analysis is carried out on Landsat time series images at pixel level to generate pattern features.Then,all features are split into"feature blocks",and base classifiers are integrated into separate sets according to feature subspaces.Finally,classification results are produced by weighted majority vote of base classifiers.Experimental results and quantitative analysis show that this method has greatly improved the classification accuracy,compared with the traditional feature extraction and classification methods,and is robust for high dimensional data.This method can effectively overcome the influence of cloud cover,data gaps and phenological change in large area land cover classification,and has high accuracy and practicability.
To overcome problems,such as difficulty of data acquisition,complex data structure,low classification accuracy and influence of seasonal variation,existing in large area land cover classification,this paper proposes a novel land cover classification method(UniBagging)which takes annual time series Landsat data as input.This method defines a feature generation strategy for time series data,and designes an ensemble classification algorithm based on subspace,according to the characteristics of time series data.The method contains two stages.Firstly,based on specific model,regression analysis is carried out on Landsat time series images at pixel level to generate pattern features.Then,all features are split into"feature blocks",and base classifiers are integrated into separate sets according to feature subspaces.Finally,classification results are produced by weighted majority vote of base classifiers.Experimental results and quantitative analysis show that this method has greatly improved the classification accuracy,compared with the traditional feature extraction and classification methods,and is robust for high dimensional data.This method can effectively overcome the influence of cloud cover,data gaps and phenological change in large area land cover classification,and has high accuracy and practicability.
引文
[1]JUNG M,HENKEL K,HEROLD M,et al.Exploiting synergies of global land cover products for carbon cycle modeling[J].Remote Sensing of Environment,2006,101(4):534-553.
[2]ROGAN J,BUMBARGER N,KULAKOWSKI D,et al.Improving forest type discrimination with mixed lifeform classes using fuzzy classification thresholds informed by field observations[J].Canadian Journal of Remote Sensing,2010,36(6):699-708.
[3]COHEN W B,GOWARD S N.Landsat’s role in ecological applications of remote sensing[J].BioScience,2004,54(6):535-545.
[4]WOODCOCK C E,ALLEN R G.Free access to Landsat imagery[J].Science,2008,320(5879):1011-1011.
[5]WULDER M A,WHITE J C,LOVELAND T R,et al.The global Landsat archive:status,consolidation,and direction[J].Remote Sensing of Environment,2016:271-283.
[6]XIE H,LUO X,XU X,et al.Evaluation of Landsat 8 OLI imagery for unsupervised inland water extraction[J].International Journal of Remote Sensing,2016,37(8):1826-1844.
[7]赵英时.遥感应用分析原理与方法[M].北京:科学出版社,2013:34-41.
[8]GOMEZ C,WHITE J C,WULDER M A,et al.Optical remotely sensed time series data for land cover classification:a review[J].ISPRS Journal of Photogrammetry and Remote Sensing,2016:55-72.
[9]ZHU Z,WOODCOCK C E.Continuous change detection and classification of land cover using all available Landsat data[J].Remote sensing of Environment,2014,144:152-171.
[10]JULIEN Y,SOBRINO J A,JIMENEZMUNOZ J C,et al.Land use classification from multitemporal Landsat imagery using the yearly land cover dynamics(YLCD)method[J].International Journal of Applied Earth Observation and Geoinformation,2011,13(5):711-720.
[11]SEXTON J O,URBAN D L,DONOHUE M J,et al.Long-term land cover dynamics by multi-temporal classification across the Landsat-5record[J].Remote Sensing of Environment,2013:246-258.
[12]LI M,IM J,BEIER C M,et al.Machine learning approaches for forest classification and change analysis using multitemporal Landsat TM images over Huntington Wildlife Forest[J].GIScience &Remote Sensing,2013.
[13]HANSEN M C,LOVELAND T R,Loveland T R.A review of large area monitoring of land cover change using Landsat data[J].Remote Sensing of Environment,2012:66-74.
[14]CHEN Y,ZHAO X,JIA X,et al.Spectral-spatial classification of hyperspectral data based on deep belief network[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2015,8(6):2381-2392.
[15]CHEN Y,JIANG H,LI C,et al.Deep feature extraction and classification of hyperspectral images based on convolutional neural networks[J].IEEE Transactions on Geoscience and Remote Sensing,2016,54(10):6232-6251.
[16]KEMKER R,KANAN C.Self-taught feature learning for hyperspectral image classification[J].IEEE Transactions on Geoscience &Remote Sensing,2017,55(5):1-13.
[17]SIMON J L,BRETAGNON P,CHAPRONT J,et al.Numerical expressions for precession formulae and mean elements for the Moon and the planets[J].Astronomy and Astrophysics,1994,282(2):663-683.
[18]HOMER C G,DEWITZ J,YANG L,et al.Completion of the 2011national 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.
[2]ROGAN J,BUMBARGER N,KULAKOWSKI D,et al.Improving forest type discrimination with mixed lifeform classes using fuzzy classification thresholds informed by field observations[J].Canadian Journal of Remote Sensing,2010,36(6):699-708.
[3]COHEN W B,GOWARD S N.Landsat’s role in ecological applications of remote sensing[J].BioScience,2004,54(6):535-545.
[4]WOODCOCK C E,ALLEN R G.Free access to Landsat imagery[J].Science,2008,320(5879):1011-1011.
[5]WULDER M A,WHITE J C,LOVELAND T R,et al.The global Landsat archive:status,consolidation,and direction[J].Remote Sensing of Environment,2016:271-283.
[6]XIE H,LUO X,XU X,et al.Evaluation of Landsat 8 OLI imagery for unsupervised inland water extraction[J].International Journal of Remote Sensing,2016,37(8):1826-1844.
[7]赵英时.遥感应用分析原理与方法[M].北京:科学出版社,2013:34-41.
[8]GOMEZ C,WHITE J C,WULDER M A,et al.Optical remotely sensed time series data for land cover classification:a review[J].ISPRS Journal of Photogrammetry and Remote Sensing,2016:55-72.
[9]ZHU Z,WOODCOCK C E.Continuous change detection and classification of land cover using all available Landsat data[J].Remote sensing of Environment,2014,144:152-171.
[10]JULIEN Y,SOBRINO J A,JIMENEZMUNOZ J C,et al.Land use classification from multitemporal Landsat imagery using the yearly land cover dynamics(YLCD)method[J].International Journal of Applied Earth Observation and Geoinformation,2011,13(5):711-720.
[11]SEXTON J O,URBAN D L,DONOHUE M J,et al.Long-term land cover dynamics by multi-temporal classification across the Landsat-5record[J].Remote Sensing of Environment,2013:246-258.
[12]LI M,IM J,BEIER C M,et al.Machine learning approaches for forest classification and change analysis using multitemporal Landsat TM images over Huntington Wildlife Forest[J].GIScience &Remote Sensing,2013.
[13]HANSEN M C,LOVELAND T R,Loveland T R.A review of large area monitoring of land cover change using Landsat data[J].Remote Sensing of Environment,2012:66-74.
[14]CHEN Y,ZHAO X,JIA X,et al.Spectral-spatial classification of hyperspectral data based on deep belief network[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2015,8(6):2381-2392.
[15]CHEN Y,JIANG H,LI C,et al.Deep feature extraction and classification of hyperspectral images based on convolutional neural networks[J].IEEE Transactions on Geoscience and Remote Sensing,2016,54(10):6232-6251.
[16]KEMKER R,KANAN C.Self-taught feature learning for hyperspectral image classification[J].IEEE Transactions on Geoscience &Remote Sensing,2017,55(5):1-13.
[17]SIMON J L,BRETAGNON P,CHAPRONT J,et al.Numerical expressions for precession formulae and mean elements for the Moon and the planets[J].Astronomy and Astrophysics,1994,282(2):663-683.
[18]HOMER C G,DEWITZ J,YANG L,et al.Completion of the 2011national 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.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |