基于機器視覺的高速公路車道偏離警告系統(tǒng)的研究.pdf

1、合肥工業(yè)大學碩士學位論文基于機器視覺的高速公路車道偏離警告系統(tǒng)的研究姓名：羅才彬申請學位級別：碩士專業(yè)：信號與信息處理指導(dǎo)教師：胡慶新201104ResearchonLaneDepartureWarningSystemofSpeedwayBasedonMachineVisionAbstractNowadayswiththerapiddevelopmentofsocialeconomytheincreasinglyexplosionina

2、utoinventytrafficaccidentshappenmefrequently.Themainreasonisthatthevehicledeviatesfromthelanecollideswiththecaronthesidelanewhendriversareoutofconsciousness.Asaresulttheauxiliarysystemofintelligentdrivingisneededwhichcan

3、avoidlanedepartureaccidentsprovidesafedriving.Lanedeparturewarningsystemisoneoftheresearchesintheauxiliarysystemofintelligentdrivingtheextractionrecognitionoflanemarklinetakesanimptantroleinthissystem.Thisdissertationhas

4、adeepresearchbasedonmonocularvisioninthejudgmentofdrivewaytypethealgithmofsubsectionextraction.Tobeginwithindertoeliminatetheirrelevantinfmationenhancethetestabilityofusefulinfmationthedrivewayimagesequencescollectedbyca

5、merawerepreprocessed.Thenadivisionofthepreprocessedimagewasmadetodistinguishtheroadareaanimageonlycontainsdrivewaymarklinewasobtained.Thedrivewaymarklineisrefinedtoreducetheamountofdata.Theprinciplesofextractionstraightl

6、ineareanalyzedbytheleastsquaresHoughtransfmthemedianintercept.Theoptimalalgithmofextractionisthenedbyexperimentsinthisdissertation.DecomposedthedrivewaymarklineintotwopartULMLLManalyzedtheiracteristics.AdoptedHoughtransf

7、mcomputingtheparametersofLLMleastsquarecomputingtheparameterofULMsummedupthejudgmentcriterionofdrivewaytypeaccdingtotherelationshipbetweenslopes.Withalargenumberofexperimentsinstraightleftrightcnerdrivewayimagethereliabi

8、lityofthecriterionisverified.Thedrivewaymarklinewasextractedbysegmentaccdingtothejudgmentofdrivewaytype.IfthelaneisstraighttheaverageofparametersofLLMULMwasusedastheparametersoflane.Ifthelaneisaturnaparabolicmodelwasesta