预览加载中，请您耐心等待几秒...

遥感图像云检测的多尺度融合分割网络方法.docx

预览
1/2
2/2

在线预览结束，喜欢就下载吧，查找使用更方便

下载提示 文本预览

如果您无法下载资料，请参考说明：

1、部分资料下载需要金币，请确保您的账户上有足够的金币

2、已购买过的文档，再次下载不重复扣费

3、资料包下载后请先用软件解压，在使用对应软件打开

遥感图像云检测的多尺度融合分割网络方法 Abstract Clouddetectionisacrucialtaskinremotesensingimageanalysis,asthepresenceofcloudscansignificantlyaffecttheinterpretationandanalysisofthecollecteddata.Inthispaper,weproposeamulti-scalefusionsegmentationnetworkmethodforclouddetectioninremotesensingimages.Theproposedmethodcombinesmultiplescalesofimageinformationtoenhancetheaccuracyandrobustnessofclouddetection.Adeepneuralnetworkistrainedusingalarge-scaledatasetofannotatedremotesensingimagestoenableautomaticclouddetection. 1.Introduction Remotesensingimagesprovidevaluableinformationforvariousapplicationssuchasenvironmentalmonitoring,disastermanagement,andurbanplanning.However,thepresenceofcloudsintheseimagescansignificantlydegradethequalityofthecollecteddata.Therefore,accurateclouddetectionisessentialfortheeffectiveanalysisandinterpretationofremotesensingimages.Variousapproacheshavebeenproposedforclouddetection,includingthresholding,texture-basedmethods,andmachinelearning-basedmethods.Inrecentyears,deeplearningtechniqueshaveshownpromisingresultsinremotesensingimageanalysis,includingclouddetection.Inthispaper,weproposeamulti-scalefusionsegmentationnetworkmethodtoaddressthechallengesinclouddetection. 2.Methodology Theproposedmethodconsistsoftwomaincomponents:multi-scalefusionandsegmentationnetwork.Themulti-scalefusioncomponentaimstoenhancetherepresentationoftheinputimageatdifferentscalesbycombiningthefeaturesextractedfrommultiplescales.Thisisachievedbyusingdilatedconvolutionsandskipconnectionstocapturebothlocalandglobalcontextinformation.Thesegmentationnetworkcomponentisresponsibleforclassifyingeachpixelintheimageintocloudornon-cloudcategories.Thiscomponentisbasedonafullyconvolutionalnetworkarchitecture,whichenablesefficientandaccuratepixel-wiseclassification. 3.ExperimentalSetup Toevaluatetheperformanceoftheproposedmethod,weconductedexperimentsonalarge-scaledatasetofremotesensingimages.Thedatasetconsistsofannotatedimageswithpixel-levellabelsindicatingthepresenceorabsenceofclouds.Theproposedmetho