欢迎来到天天文库
浏览记录
ID:36803729
大小:2.22 MB
页数:50页
时间:2019-05-15
《遥感影像分割边界平滑方法研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、摘要遥感影像含有丰富的地物信息,通过对分割遥感影像所得的对象进行边界平滑,得到相对真实的对象边界,是提高遥感信息提取精度的重要手段。光滑后的对象边界更加接近实际地物的边界,更好的反映人眼的视觉要求,对后续的遥感影像分类和遥感信息提取至关重要。本文针对对象边界固有的复杂性和边界特征点的分布特点,提出了一套完整高效的平滑方法,该方法的主要流程是:首先对分割后所得到的每个对象的边界进行按节点拆分;然后对每个边界段进行边界去噪、断点筛选等预处理;再通过DP算法提取代表边界信息的特征点;最后使用三次B样条拟合所得到的特征点,完成边界平滑。本文的主要
2、创新点有:11)针对多个相邻对象之间边界平滑问题,提出了一种有效的解决方法;12)设计并实现了一种提取断点的方法,大大减少了边界点数量,提高了特征点检测速度和精度;(3)实现了特征点检测的DP算法。本文用实际遥感影像对进行分割后的对象边界进行了边界提取、边界拆分、预处理及特征点提取和三次B样条拟合等实验,实验表明本文方法能以可接受的速度和精度对原对象边界进行光滑处理。关键词:遥感影像,对象边界,特征点,曲线拟合AbstractRemotesensingimagecontainawealthofinformationofsurfacefea
3、tures,throughremotosensingimagesegmentationofobjectswithsmoothbordersandboundariesarcrelativelyrealobjectistoimprovetheprecisionofremotesensinganimportantmeansofinformationextraction.Smoothobjectboundaryafterboundaryclosertotheactualsurfacofeatures,betterreflectthehumaney
4、e'svisualrequirementsofthefollow-upofremotesensingimageclassificationandextractionofessentialinformation.Inthispaper,theinherentcomplexityoftheobjectboundary,andboundaryfeaturepointsdistribution,proposedacompletesmoothingefficientmethodforthemainprocessis:afterthefirstspl
5、itbytheborderofeachobjectaccordingtothenodesplit;thencarriedoutforeachbordersegmentboundarydenoising,pretreatmentscreeningbreakpoint;thenextractedthroughtheDPalgorithmoffeaturepointinformationonbehalfoftheborder;FinallycubicB—splinefiUingfeaturepointsobtained,completewith
6、smoothborders.Themaininnovationofthispaperare"(1)forsmoothingOVertheborderissuebetweenadjacentobjects,proposesaneffectivesolution;(2)desi印andimplementamethodofextractingbreak,greatlyreducingtheboundarypoints,andimprovedthespeedandaccuracyoffeaturepondetection;(3)implement
7、ationofthefeaturepoiIltdetectionoftheDPalgorithm.Inthispaper,theactualsegmentationofremotesensingimageoftheobjectboundaryaRerboundaryextraction,boundaryseparation,pretreatment,andfeatureextractionandcubicB-splinefittingsuchexperimentsshowedthatthismethodcanbeanacceptables
8、peedandaccuracyoftheoriginalobjectboundarysmoothing.Keyword:remotesensingimage,objectboundary,fe
此文档下载收益归作者所有