基于PUL算法及高分辨率WorldView影像的城市不透水面提取
|
摘要
准确提取城市不透水面对生态环境、水热循环及热岛效应等研究具有重要意义。该文利用WorldView高分辨遥感影像,提出基于PUL(Positive and Unlabeled Learning)算法的高分辨率影像城市不透水面提取方法,该方法不需要负样本数据,只需少量的正样本和未标记样本即可训练分类模型。结果显示,PUL算法的提取结果优于一类支持向量机(OCSVM)以及最大熵(MAXENT)模型。使用不同正样本量时,PUL的提取结果总体精度和kappa系数均优于OCSVM和MAXENT,最高总体精度为91.27%,最高kappa系数可达0.8255,可快速、有效地从高分辨率遥感影像中提取不透水面。
Correctly mapping the urban impervious surfaces is important in the studies of urban environment,hydrothermal cycle and the heat island effect.In this study,the positive and unlabeled learning(PUL)algorithm was investigated,which trained a classifier on positive and unlabeled data,to map the urban impervious surfaces based on the high-resolution WorldView images.Different from traditional supervised classification methods,which require both labeled positive and negative training data,the PUL algorithm requires only labeled positive and unlabeled training data.Experimental results show that the PUL algorithm outperforms the One-Class Support Vector Machine(OCSVM)and the Maximum Entropy(MAXENT)methods.For all training sample sizes,PUL consistently produces higher accuracies(i.e.overall accuracy and kappa coefficient)than the other two methods.The highest overall accuracy and kappa coefficient obtained by PUL are 91.27% and 0.8255,respectively.Thus,PUL is an efficient and promising method for extracting impervious surfaces from high-resolution remote sensing images.
Correctly mapping the urban impervious surfaces is important in the studies of urban environment,hydrothermal cycle and the heat island effect.In this study,the positive and unlabeled learning(PUL)algorithm was investigated,which trained a classifier on positive and unlabeled data,to map the urban impervious surfaces based on the high-resolution WorldView images.Different from traditional supervised classification methods,which require both labeled positive and negative training data,the PUL algorithm requires only labeled positive and unlabeled training data.Experimental results show that the PUL algorithm outperforms the One-Class Support Vector Machine(OCSVM)and the Maximum Entropy(MAXENT)methods.For all training sample sizes,PUL consistently produces higher accuracies(i.e.overall accuracy and kappa coefficient)than the other two methods.The highest overall accuracy and kappa coefficient obtained by PUL are 91.27% and 0.8255,respectively.Thus,PUL is an efficient and promising method for extracting impervious surfaces from high-resolution remote sensing images.
引文
[1]任金华,吴绍华,周生路,等.城市不透水面遥感研究进展[J].国土资源遥感,2012,24(4):8-15.
[2]王浩,卢善龙,吴炳方,等.不透水面遥感提取及应用研究进展[J].地球科学进展,2013,28(3):327-336
[3]李仕峰,钱乐祥,王瑾.基于陆地卫星TM/ETM+改进的温湿指数及其对不透水表面的响应[J].地理与地理信息科学,2013,29(2):112-115.
[4]刘珍环,王仰麟,彭建.不透水表面遥感监测及其应用研究进展[J].地理科学进展,2010,29(9):1143-1152.
[5]黄曦涛,李怀恩,张瑜,等.利用影像纹理和阴影信息提取城市不透水面的方法——以咸阳市为例[J].测绘通报,2016,2016(5):80-83.
[6]张俊,于庆国,侯家槐.面向对象的高分辨率影像分类与信息提取[J].遥感技术与应用,2010,25(1):112-117.
[7]李苗,臧淑英,吴长山,等.影像分割的城市不透水面信息提取[J].测绘科学,2017,42(2):84-87.
[8]LU D,TIAN H,ZHOU G,et al.Regional mapping of human settlements in southeastern China with multisensor remotely sensed data[J].Remote Sensing of Environment,2008,112(9):3668-3679.
[9]ZHANG Y,ZHANG H,LIN H.Improving the impervious surface estimation with combined use of optical and SAR remote sensing images[J].Remote Sensing of Environment,2014,141(2):155-167.
[10]HU X,WENG Q.Estimating impervious surfaces from medium spatial resolution imagery using the self-organizing map and multi-layer perceptron neural networks[J].Remote Sensing of Environment,2009,113(10):2089-2102.
[11]徐涵秋.一种快速提取不透水面的新型遥感指数[J].武汉大学学报(信息科学版),2008,33(11):1150-1153.
[12]刘波,张源,程涛,等.基于高分二号卫星影像的城市不透水面提取[J].地理信息世界,2017,24(2):103-107.
[13]李彩丽,都金康,左天惠.基于高分辨率遥感影像的不透水面信息提取方法研究[J].遥感信息,2009(5):36-40.
[14]MOHAPATRA R P,WU C.High resolution impervious surface estimation[J].Photogrammetric Engineering&Remote Sensing,2010,76(12):1329-1341.
[15]李德仁,罗晖,邵振峰.遥感技术在不透水层提取中的应用与展望[J].武汉大学学报(信息科学版),2016,41(5):569-577.
[16]程熙,沈占锋,骆剑承,等.“全域—局部”不透水面信息遥感分步提取模型[J].遥感学报,2013,17(5):1191-1205.
[17]SCHLKOPF B,PLATT J C,SHAWE-TAYLOR J,et al.Estimating the support of a high-dimensional distribution[J].Neural Computation,2001,13(7):1443-1471.
[18]RICHARD M D,LIPPMANN R P.Neural network classifiers estimate Bayesian a posteriori probabilities[J].Neural computation,1991,3(4):461-483.
[19]BHNING D.Multinomial logistic regression algorithm[J].Annals of the Institute of Statistical Mathematics,1992,44(1):197-200.
[20]LI W,GUO Q,ELKAN C.A positive and unlabeled learning algorithm for one-class classification of remote-sensing data[J].IEEE Transactions on Geoscience and Remote Sensing,2011,49(2):717-725.
[21]LI W,GUO Q.A maximum entropy approach to one-class classification of remote sensing imagery[J].International Journal of Remote Sensing,2010,31(8):2227-2235.
[22]LIU B,DAI Y,LI X,et al.Building text classifiers using positive and unlabeled examples[A].IEEE International Conference on Data Mining[C].2003.179-186.
[23]SANCHEZ-HERNANDEZ C,BOYD D S,FOODY G M.Oneclass classification for mapping a specific land-cover class:SVDD classification of fenland[J].IEEE Transactions on Geoscience and Remote Sensing,2007,45(4):1061-1073.
[24]SONG B,LI P,LI J,et al.One-class classification of remote sensing images using kernel sparse representation[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2016,9(4):1613-1623.
[25]KHAN S S,MADDEN M G.One-class classification:Taxonomy of study and review of techniques[J].The Knowledge Engineering Review,2014,29(3):345-374.
[26]TAX D M J.One-class classification:Concept-learning in the absence of counter-examples[D].Delft:Delft University of Technology,2001.65.
[27]PHILLIPS S J,ANDERSON R P,SCHAPIRE R E.Maximum entropy modeling of species geographic distributions[J].Ecological Modelling,2006,190(3):231-259.
[28]BALDECK C A,ASNER G P.Single-species detection with airborne imaging spectroscopy data:A comparison of support vector techniques[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2015,8(6):2501-2512.
[29]ELKAN C,NOTO K.Learning classifiers from only positive and unlabeled data[A].ACM SIGKDD International Conference on Knowledge Discovery and Data Mining[C].2008.213-220.
[30]CASTELLI V,COVER T M.The relative value of labeled and unlabeled samples in pattern recognition with an unknown mixing parameter[J].IEEE Transactions on Information Theory,1996,42(6):2102-2117.
[31]WAN B,GUO Q,FANG F,et al.Mapping US urban extents from MODIS data using one-class classification method[J].Remote Sensing,2015,7(8):10143-10163.
[32]LI W,GUO Q.A new accuracy assessment method for oneclass remote sensing classification[J].IEEE Transactions on Geoscience and Remote Sensing,2014,52(8):4621-4632.
[33]GUO Q,LI W,LIU D,et al.A Framework for Supervised Image Classification with Incomplete Training Samples[J].Photogrammetric Engineering&Remote Sensing,2012,78(6):595-604.
[2]王浩,卢善龙,吴炳方,等.不透水面遥感提取及应用研究进展[J].地球科学进展,2013,28(3):327-336
[3]李仕峰,钱乐祥,王瑾.基于陆地卫星TM/ETM+改进的温湿指数及其对不透水表面的响应[J].地理与地理信息科学,2013,29(2):112-115.
[4]刘珍环,王仰麟,彭建.不透水表面遥感监测及其应用研究进展[J].地理科学进展,2010,29(9):1143-1152.
[5]黄曦涛,李怀恩,张瑜,等.利用影像纹理和阴影信息提取城市不透水面的方法——以咸阳市为例[J].测绘通报,2016,2016(5):80-83.
[6]张俊,于庆国,侯家槐.面向对象的高分辨率影像分类与信息提取[J].遥感技术与应用,2010,25(1):112-117.
[7]李苗,臧淑英,吴长山,等.影像分割的城市不透水面信息提取[J].测绘科学,2017,42(2):84-87.
[8]LU D,TIAN H,ZHOU G,et al.Regional mapping of human settlements in southeastern China with multisensor remotely sensed data[J].Remote Sensing of Environment,2008,112(9):3668-3679.
[9]ZHANG Y,ZHANG H,LIN H.Improving the impervious surface estimation with combined use of optical and SAR remote sensing images[J].Remote Sensing of Environment,2014,141(2):155-167.
[10]HU X,WENG Q.Estimating impervious surfaces from medium spatial resolution imagery using the self-organizing map and multi-layer perceptron neural networks[J].Remote Sensing of Environment,2009,113(10):2089-2102.
[11]徐涵秋.一种快速提取不透水面的新型遥感指数[J].武汉大学学报(信息科学版),2008,33(11):1150-1153.
[12]刘波,张源,程涛,等.基于高分二号卫星影像的城市不透水面提取[J].地理信息世界,2017,24(2):103-107.
[13]李彩丽,都金康,左天惠.基于高分辨率遥感影像的不透水面信息提取方法研究[J].遥感信息,2009(5):36-40.
[14]MOHAPATRA R P,WU C.High resolution impervious surface estimation[J].Photogrammetric Engineering&Remote Sensing,2010,76(12):1329-1341.
[15]李德仁,罗晖,邵振峰.遥感技术在不透水层提取中的应用与展望[J].武汉大学学报(信息科学版),2016,41(5):569-577.
[16]程熙,沈占锋,骆剑承,等.“全域—局部”不透水面信息遥感分步提取模型[J].遥感学报,2013,17(5):1191-1205.
[17]SCHLKOPF B,PLATT J C,SHAWE-TAYLOR J,et al.Estimating the support of a high-dimensional distribution[J].Neural Computation,2001,13(7):1443-1471.
[18]RICHARD M D,LIPPMANN R P.Neural network classifiers estimate Bayesian a posteriori probabilities[J].Neural computation,1991,3(4):461-483.
[19]BHNING D.Multinomial logistic regression algorithm[J].Annals of the Institute of Statistical Mathematics,1992,44(1):197-200.
[20]LI W,GUO Q,ELKAN C.A positive and unlabeled learning algorithm for one-class classification of remote-sensing data[J].IEEE Transactions on Geoscience and Remote Sensing,2011,49(2):717-725.
[21]LI W,GUO Q.A maximum entropy approach to one-class classification of remote sensing imagery[J].International Journal of Remote Sensing,2010,31(8):2227-2235.
[22]LIU B,DAI Y,LI X,et al.Building text classifiers using positive and unlabeled examples[A].IEEE International Conference on Data Mining[C].2003.179-186.
[23]SANCHEZ-HERNANDEZ C,BOYD D S,FOODY G M.Oneclass classification for mapping a specific land-cover class:SVDD classification of fenland[J].IEEE Transactions on Geoscience and Remote Sensing,2007,45(4):1061-1073.
[24]SONG B,LI P,LI J,et al.One-class classification of remote sensing images using kernel sparse representation[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2016,9(4):1613-1623.
[25]KHAN S S,MADDEN M G.One-class classification:Taxonomy of study and review of techniques[J].The Knowledge Engineering Review,2014,29(3):345-374.
[26]TAX D M J.One-class classification:Concept-learning in the absence of counter-examples[D].Delft:Delft University of Technology,2001.65.
[27]PHILLIPS S J,ANDERSON R P,SCHAPIRE R E.Maximum entropy modeling of species geographic distributions[J].Ecological Modelling,2006,190(3):231-259.
[28]BALDECK C A,ASNER G P.Single-species detection with airborne imaging spectroscopy data:A comparison of support vector techniques[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2015,8(6):2501-2512.
[29]ELKAN C,NOTO K.Learning classifiers from only positive and unlabeled data[A].ACM SIGKDD International Conference on Knowledge Discovery and Data Mining[C].2008.213-220.
[30]CASTELLI V,COVER T M.The relative value of labeled and unlabeled samples in pattern recognition with an unknown mixing parameter[J].IEEE Transactions on Information Theory,1996,42(6):2102-2117.
[31]WAN B,GUO Q,FANG F,et al.Mapping US urban extents from MODIS data using one-class classification method[J].Remote Sensing,2015,7(8):10143-10163.
[32]LI W,GUO Q.A new accuracy assessment method for oneclass remote sensing classification[J].IEEE Transactions on Geoscience and Remote Sensing,2014,52(8):4621-4632.
[33]GUO Q,LI W,LIU D,et al.A Framework for Supervised Image Classification with Incomplete Training Samples[J].Photogrammetric Engineering&Remote Sensing,2012,78(6):595-604.