基于视觉的道路检测技术分析

《基于视觉的道路检测技术分析》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、业爿L蔓L』—厶—竺—蜓—上—竺—也—盈—=堑——————』—旦jL工』生j二—IABSTRAC_rABSTRACT：Currently,researchonintelligentvehicletechnologiesisbecominghotinmanyfields，andhasattractedmuchinterestfromtheresearchscientistsOfthesetechnologiesvisionbasedvehiclenavigationisoneofthemostsignificantareas，anditmainlyconcen

2、【ratesolltheproblemofhowtodetectroadlaneinformation"iqaeworkofthisthesisisgivenasfolfowsFirstly，thebackgroundandsignificanceoftheroadlanedetectiontechnologyusingcomputervisionisintroduced，aswellastheadvancesofintelligentvehicleresearchathomeandabroad，andtwoconventionalroadlanedetec

3、tionmethods，namelymodelbasedmethodandfeaturebasedmethod．arestudiedindetailSecondly,imagepre—processingalgorithmsarepresentedforthevehiclecameravideo，includingcolorimagegrayconversion，histogramequaliTation，andmedianfiltertorealizelinagenoisefilteringAself-defineddifferenceoperatoris

4、thenproposedtoextracttheimagecdge，bythesumofapixelanditseightadjacentpixclswithdifferentweightsExperimentalresultsshowthattheself_defineddifllcrenceoperatorisabletoachievebetterpertbrmancetoobtainimageedgefeaturesMeanwhile，regiongrowingalgorithmispresentedheretoextracttheregionalfe

5、aturesofanimagebyregiondlvjshinfhen，analgorithmintegratededgeandregionfeaturesisproposedtoimplementedgesegmentation，usingthecomplementarityoftheirfeatures，andthenthelineparametersareobtainedusingHoughtransformationUsingtheslopeolroadlanesandtheintersectionangleoftwolanesinanimageth

6、elanescanthusberecognizedFinally,plentyofexperimentswerecarriedoutundervariedilluminationandcomplicmedenvironmentconditionsTheseresultssho'wthatthealgorithmsproposedinthisthesiscanbeusedtoidentifytheroadlanesfromthevideolastlyandrobustlyKEYWORDS：vehiclecanleFavideo：roadlanedetectio

7、n；self-delineddifferenceoperator；regiongrowingmethod；edgeandregionalfeatureintegration；HoughtransformationCI。ASSNO：丁P39l4：】E噩至堑厶生熊±堂位丝塞————————旦』目录q·文摘要一⋯⋯⋯iiiABSTRACr⋯i、，1日l?；一⋯1l课题研究背景与意义⋯l12视觉导航研究现状⋯I121旧外研究现状12,21日内研究现状⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯313本文研究内容42道路图像特点及议别方法⋯621娈化环境下的道路图像特点⋯62螭1。视觉的

8、道蹄U瑚J方法1122l壮f道路模型识别方法⋯122