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
基于单目视觉的夜间车道线和前方车辆检测方法研究的中期报告(ThisisamachinetranslationfromChinesetoEnglish.Itmaycontainerrorsorinconsistencies.)MidtermReportonResearchofNighttimeLaneDetectionandVehicleDetectionMethodbasedonMonocularVisionIntroductionWiththedevelopmentofadvanceddriverassistancesystems(ADAS)andautonomousdrivingtechnologies,theresearchonperceptionandidentificationofthesurroundingenvironmentbecomesmoreandmorecritical.Amongthem,thevision-basedsystemiswidelyusedforsolvingthisclassofproblems.MonocularvisionisoneoftheessentialmethodsfortheperceptionanddetectionofthesurroundingenvironmentinADASandautonomousvehicles.Itplaysavitalroleinobjectdetection,lanekeeping,pedestriandetection,andtrafficsignrecognition,etc.However,thenightenvironmentismorecomplexthanthedaytimeenvironment.Inthenight,thelightingisinsufficient,thecontrastoftheimageisreduced,andthenoiseissignificantlyincreasedintheimage,whichmakesitquitechallengingtodetectandperceivetheobjectsthroughthecamera.Consequently,detectingthelaneandvehicleinalowilluminationenvironmentbecomesasignificantchallenge.Therefore,studyingthenight-timevisionhasattractedconsiderableattentioninthefieldofcomputervisioninrecentyears.Inthisproject,weaimtodesignaneffectivemonocularvision-basedmethodfornighttimelanedetectionandvehicledetection.Inthismidtermreport,wepresenttheprogressoftheresearchthatwehavecompletedsofar.MethodsNight-timeLaneDetectionTheproposedlanedetectionmethodhasthreemaincomponents:Preprocessing,Lanerecognition,andPostprocessing.Preprocessing:Theimagecapturedbythecameraispreprocessed,whichincludesimageenhancement,imagedistortioncorrection,andregionofinterest(ROI)extraction.First,thehistogramequalizationalgorithmisusedtoenhancethebrightnesscontrast,whichcansignificantlyincreasetheimage'svisibility.Second,weapplytheDistortioncorrectionbasedonthecamera'sintrinsicandextrinsicparameters,whichcanrectifythecurvedimagecausedbythelensdistortion.Next,weidentifytheregionofinterest(ROI)oftheimage,whichhelpsustoreducethecomputationloadofthealgori