预览加载中,请您耐心等待几秒...
1/2
2/2

在线预览结束,喜欢就下载吧,查找使用更方便

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

一种实时鲁棒的直线提取技术 Abstract: Lineextractionisafundamentaltaskincomputervisionandrobotnavigation.Inthispaper,weproposeareal-timeandrobustlineextractiontechniquethatcanhandlevarioustypesofimagenoiseandclutter.Theproposedtechniqueconsistsofthreemainsteps:pre-processing,linecandidategeneration,andlinerefinement.Weevaluateourtechniqueondifferentdatasetsandcompareitwithstate-of-the-artmethods.Theresultsshowthatourmethodachieveshigheraccuracywithafasterprocessingtime. Introduction: Extractinglinesfromimagesisacrucialtaskinmanycomputervisionandroboticsapplications.Forexample,inautonomousnavigation,robotsneedtoidentifythelinesontheroadtocontroltheirmotion.Lineextractioncanalsobeusedinobjectrecognition,wherelinesareusedtodescribetheshapeofanobject.However,lineextractionfromimagesischallengingduetovarioustypesofnoiseandclutter. Therearemanytechniquesforlineextraction,suchasHoughtransform,edgedetection,andtemplatematching.However,thesetechniquessufferfromvariouslimitationssuchassensitivitytonoise,lossofdetail,andhighcomputationalcost.Moreover,thesetechniquesmaynotbesuitableforreal-timeapplications. Toaddressthesechallenges,weproposeareal-timeandrobustlineextractiontechniquethatcanhandlevarioustypesofimagenoiseandclutter.Theproposedtechniqueconsistsofthreemainsteps:pre-processing,linecandidategeneration,andlinerefinement.Inthepre-processingstep,weapplyanoisereductionfiltertotheinputimagetoremovethenoiseandenhancetheedgedetails.Inthelinecandidategenerationstep,weapplyalinesegmentdetectortodetectthelinesegmentsinthepre-processedimage.Finally,inthelinerefinementstep,weapplyaclusteringalgorithmtogroupthedetectedlinesegmentsintocoherentlines. ProposedTechnique: Pre-processing: Theinputimageisfirstpre-processedtoenhanceitsedgedetailsandremovethenoise.WeuseaGaussianfiltertosmooththeimageandsuppressthenoise.Wethenapplyahigh-passfiltertoenhancetheedgedetails.Thehigh-passfilterisimplementedbysubtractingtheGaussian-smoothedimagefromtheoriginalimage.Finally,weapplyathresholdtoobtainabinaryedgema