摄像机标定方法分析

《摄像机标定方法分析》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、东北大学硕士学位论文摘要关键词：视频监控；摄像机标定；消失点；消失线：遗传算法；奇异值分解：一IV—东北大学硕士学位论文AbstractStudyonCameraCalibrationMethodsAbstractAsthemodemcomputertechnologygrowingfastly，anemergingtechnology，ComputerVision，hasgraduallyformed．ItmainlyUSeScomputertoextractinformationsfromvideosorphotosforsimulatingpeop

2、le’Svisualcapabilities，thenprocessesandunderstandingthem，atlast，theseinformationsCanbeusedforreal—timemonitoringandcontrolling．IntelligentvisualsurveillanceisoneofthemostimportantapplicationsinComputerVision，itusescamerascontrolledbycomputertoreplacepeople，undertakingthemonitorin

3、gtasks．Itanalysisesandprocessesasequenceofvideoframestoimplementmonitoring．However，cameracalibrationisaprimaryandvitaltopicinintelligentvisualsurveillance．Thispaperintroducesthebackgroundandbasictheoryofcameracalibration，thendiscussesflu'eemethodsofcameracalibrationandmakesrelate

4、dexperiments．Themainjobofthispaperisasfollows：firstly，weintroducetheexsistingcalibrationapproachbasedon2Dcalibrationpattem，itgetsthecameraintrinsicandextrinsicparametersbythepropertyofvanishingpoint，thenoptimizesandrefinestheintialvaluesbyGuass．Newton．Afterthat，fortheproblemofthe

5、abovemethodCannotcalibratingsomespecialkindofpictures，weproposeanewcalibrationalgorithmtoimprovetheabove—mentionedapproach，accordingtotheresultsoftheexperiments，thenewmethodCanovercometheproblemwhichtheoldonehas．Secondly，becausetheclassicnon-linearcameracalibrationmethodhascomput

6、ationalcomplexityandpoorconvergenceandisEasytofallintolocaloptimumandcallnotguaranteetherotationmatrixtobetheorthogonalmatrix，weproposeatwo-stagecalibrationmethodbasedonGeneticalgorithm，anddesignthealgorithmflow，theexperimentsverifythatthismethodCansolvetheproblemswhichmentioneda

7、bove．andiSmoreaccuratethanthemethodinchapter3．Thirdly，therequirementofthecameracalibrationintheintelligentvisualsurveillanceisthatthecalibrationobjectisassimpleandcommonaspossibleandthecalibrationmethodisasfastaspossible，butthecalibrationapproachbasedon2DcalibrationpattemCannotme

8、ettheabovedemands，SOforthisproblem，wepre