Title:Surfacecoalmineareamonitoringusingmulti-temporalhigh-resolutionsatelliteimagery利用多时段高分辨率卫星影像监控露天煤矿地区来源:国际煤炭地质学杂志2011年见刊作者:NurayDemirel等土耳其安卡拉中东理工大学采矿工程系数据库:ELSEVIERAbstract:Surfaceminingactivities,exploitationoforeandstrippinganddumpingoverburden,causechangesonthelandcoverandlanduseoftheminearea.露天采矿活动,矿石的开采和表土的剥离和倾倒,造成矿区土地覆盖和土地使用变化。Sustainableminingrequirescontinuousmonitoringofthesechangestoidentifythelong-termimpactsofminingonenvironmentandlandcovertoprovideessentialsafetymeasures.可持续开采需要连续监测这些变化来确定开采对环境和土地覆盖的长期影响提供必要的安全措施。Inthissense,digitalimageclassificationprovidesapowerfultooltoobtainarigorousdataandhencediminishestheessenceoftime-consumingandcostlyfieldmeasurements.在这个意义上说,数字图像分类提供了一个强大的工具来获得严格的数据而且因此减少实地测量的时间和费用。Therearevariousimageclassificationtechniques,servingdifferentfeaturesfordifferentpurposes,andtheSupportVectorMachine(SVM)classificationmethodbasedonstatisticalmachinelearningtheorystandsouttobeaneffectiveandaccurateimageclassificationtechniqueamongthem.各种不同的图像分类技术,服务于不同目有不同功能,支持向量机(SVM)的分类方法是基于统计机器学习理论,代表一个有效和准确的图像分类技术。ThisresearchstudyinvestigatestheuseofSVMclassificationmethodsforidentifying,quantifying,andanalyzingthespatialresponseoflandscapeduetosurfaceminingactivitiesinGoynukopencastmine,Turkey,fromyear2004to2008.本研究探讨土耳其高纳克的露天矿山从2004年至2008年,使用SVM分类方法识别,量化,分析由于地表采矿活动产生的空间响应景观。Theresearchalgorithmessentiallyentails(i)acquiringdata,(ii)pre-processingthedata,(iii)performingimageclassification,(iv)accuracyassessmentandchangedetection,and(v)analysisofresults.TheresultsshowedthatSVMclassificationmethodcaneffectivelybeutilizedforhighspatialresolutionmultispectralsatelliteimagesforidentifyingthechangesinsurfacecoalminearea.研究算法本质上需要(一)数据采集,(二)预处理数据,(三)进行图像分类,(四)准确评估和变化检测(五)分析和结果。结果表明,支持向量机分类方法可以有效地被利用在高空间分辨率多光谱卫星图像确定表面煤矿矿区范围内的变化上。Keyword:Remotesensing:遥感Imageclassification:影响分类SupportVectorMachine(SVM):支持向量机Changedetection:变化监测Surfacecoalmine:露天煤矿Landusechange利用变化1.Introduction介绍Anincreasedfocusonsustainabledevelopmentcompelsminingengineerstoprioritizeenvironmentalresponsibilitywhichrequireseffectivemanagementofsurfaceminingoperations.随着增加对可持续发展的焦点,迫使采矿工程师们优先考虑对环境的要求露天采矿作业的有效管理的责任。Inanefforttoaccomplishthis,intensivemonitoringofsurfacecoalmineareaanddisturbedlandisessentialandearthobservationtechnologycanplayacentralroleinthisprocess.要努力做到这一点,露天煤矿区和扰动土地的监测是必不可少的,对地观测技术可以在这一进程中发挥核心作用。Inadditiontothis,monitoringandmappingofsurfacemineareaprovidesdecisionmakerswithavaluabletoolfornumerousreasons,suchas,siteselecting,determiningabandonedandunreclaimedsurfacemineland,anddeterminingthechangesonlanduseinducedbyminingactivities.除了这个,为众多原因,监测和映射露天矿区提供有价值的工具,如决策者,选址,决定放弃和荒地露天矿土地,并确定采矿活动引起的土地利用变化。Insurfacecoalmininglargevolumesofwastematerialareexcavatedandremovedfromoneplacetoanothercausingcontinuouschangeintopographywithtime.在露天的煤炭开采废料从一个地方到另一个的大量挖掘和移除造成地形随时间连续变化。Largeamountsofrockandarestrippedandstoredwhichconsequentlycreateshugeholesandhugepilesinthelandscapeandleaves“scars”ontheEarthsurface.大量岩石和剥离和存储造成巨大的洞和痔,并在地球表面上留下“伤疤”。Removingtopsoil,asessentialsteptoaccessandtostripoverburdenrockmaterial,resultsindisplacementofvegetablesoilandagriculturalland,changeinbiodiversityduetodeforestation,disturbanceoflocalwaterresourcesduetoreductionincatchmentsareasanddestructionofstream.移动顶部的土壤,作为必不可少的步骤来接入和剥离表土,导致在菜地和农业用地的位移,导致砍伐森林生物多样性的变化,导致当地水资源的干扰减少集水区和溪流的毁坏。Minimizingofthesefootprintsofminingactivitiesandshiftcurrentminingpracticestoasustainableminingarekeydriverstowardsminelandmonitoring.尽量减少采矿活动和位移的脚步,改变目前的开采,实践可持续发展的开采是对矿区土地监测的关键驱动力。Conventionaltechniquesinminelandmonitoring,suchas,topographicmeasurementsandphotogrammetricstudiesaretimeconsumingandlaborintensive,therefore,arenotefficientlyusedforalarge-scalesurfaceminearea(Andersonetal.,1977).常规技术在矿区土地监测,如地形测量和摄影测量研究是费时并且是劳动密集型的,因此这些技术不能有效地用于大型露天矿区(Anderson等,1977)。Inthissense,theuseofremotelysensedsatellitedatahasbeenwidelyappliedtoprovideacost-effectivemeansofsupplyinglandcovermappingoveralargegeographicregionsinaquickertime.在这个意义上说,远程遥感卫星数据已被广泛应用,他提供在更快的地理区域土地覆盖映像提供具有成本效益的手段。Moreover,increasingnumberofsensorsandsatellitesresultsineconomicallyaffordablespatialup-to-dateinformationabouttheprocessestakingplaceonthelandbytheuseofimageclassificationalgorithms.此外,增加经济适用有关的空间最新信息传感器和卫星结果代替了土地上所使用的图像分类算法的进程。Remotesensingdatahasbeenextensivelyusedbyminingindustryformineralexplorationandmodeling,andenvironmentalmonitoringpurposesfordecades.遥感数据已经被广泛的应用在矿业开采和建模,同时也是近十年环境监测的目标。Therehavebeennumerousresearchesincorporatingminemonitoringproblemswiththevariousdigitalimageclassificationtechniqueswhichrangefromunsupervisedclassificationtosupervisedclassification,parametricclassificationtonon-parametricclassification,perpixelclassificationtosub-pixelclassificationtocharacterizethelandscapeoflargeregionsbymeansofassigningpixelsofdifferentspectralvaluestoclasses.已经有无数的研究结合各种数字图像分类技术,范围从非监督分类到监督分类,从非参数分类到参数分类法,每个像素分类的子像素的分类手段的大区域的景观特征的矿井监控问题分配不同的谱值的像素类。Variousdigitalimageclassificationtechniquescurrentlyexistwiththeirownstrengthsandlimitations.现在不同的数字影像分类技术有不同的长处和局限性。