顯微圖象特征匹配與三維顯示初步研究.pdf

1、大連理工大學(xué)碩士學(xué)位論文顯微圖象特征匹配與三維顯示初步研究姓名：宋展申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：機(jī)械制造及其自動(dòng)化指導(dǎo)教師：劉沖20030301顯微圖象特征匹配與三維顯示柳步研究AbstractMicrostereovision(MSV)istheapplicationofstereovisiontechniquesinmicrocosmicdomainStereomicroscopeandCCDsensorsmakethemaincomp

2、onentsofaMSVsystem，Inordertorealizethe3Dreconstructionandmeasurementofmicrocosmicobjectsandscenes，thestereovisionmethodiSusedusuallyInthispaper，suchproblemsasimagepro—process，characterextraction，stereomatching，depthrecov

3、ering，displayofreconstructionresultarestudiedAccordingtothecharactersofmicroscopicimages，severalmethodsareusedtodecreasethenoiseofimageandtoincreasetheimagecontrastinproprocessTheobjectiveareaandthebackgroundareaaresepar

4、atedwellandthequalityofimageiSincreasedobviouslyaftcrpreprocessBasedonthebasictheoryofcharacterextraction，theedgedetectionoperatoriSchosentoacttheroleofspecialpointsextractionoperator。Inordertochooseaneffectiveoperatorse

5、veraloperatorssuchasSobel，Laplacian，LoGandlocalentropyofedgedetectionareusedAnewalgorithmbasedOnedgesamplingandentropycurveanalysisisbroughtforwardthroughtheimprovementofthelocalentropyalgorithmTheoperatorsofLoGandlocale

6、ntropyareselectedtoextractthespecialpointsofstereoimageTheprimaryanalysisisdonetotheextractionresultStereoatchingistheemphasisofthispaperSeveralmatchingelementssuchasgrayscale，specialpointsandthemixofthemareadopted。Thema

7、ximumCROSScorrelationmethodischosenasthelikelihoodmeasurementThedisparitymapofaresistanceisgainedbythismethodTorecoverthe3DdepthinformationastereovisionmodelgOcalledWeakDisparityVisionModel(WDVM)iSintroducedAndtheprocess

8、ofcalibratingusedcalibratingtemplateislistedOpenGLisselecteda5thedevelopingenvironmenththestepofsteTeodisplayTheprocessingof3Ddiscretedatamainlyincludesthefollowingsteps：datapro—process，triangulationof3Ddiscretepoints，tr

9、iangularBernstein—BeziercurveinterpolationSomeworksof3Dpointspre—processandDelaunaytriangulation(Dr)of2DdiscretepointsaredoneinthispaperAnewDTmethodisbroughtforwardthroughtheimprovementofdatastructureandprocessingflowbas

10、edonthealgorithmofCline—RenkaSeveralmatchingmethodsareusedinthereconstructionexperimentsofaplastictempletandaresistance’SsurfaceInthereconstructionexperimentofmicrochannel’Scrosssectiontwomethodswereused：mixedcharactersa

11、ndedgemarksandthereconstructionerrorofdepthreaches28％and12％respectivelyTheresultofexperimentsshowsthatitiSpossibleteconstructandmeasurethemicrocosmicobjectthroughstereomatchingmethodButthematchingelementsandoperatorsshou