基于辅助轮廓线的医学图像三维表面重建技术研究

河南科技大学硕士学位论文基于辅助轮廓线的医学图像三维表面重建技术研究姓名：韩利华申请学位级别：硕士专业：计算机应用技术指导教师：黎蔚;陈家新20080501IIIMATLABIIISubject:ResearchontheTechnologyofMedicalImage3DSurfaceReconstructionBasedauxiliarycontourSpecialty:ComputerApplicationsTechnologyName:HANLihuaSupervisor:AssociateProfessorLIWeiProfessorCHENJiaxinABSTRACTMedicalimage3Dsurfacereconstructioncanprovidelivelyeffectdiagramsfordoctors.These3Dgraphicscangreatlyhelpthedoctoranalysisthepatients'disease.However,theadjacenttwointerbeddeddistanceofCTimageobtainedbymedicalequipmentismorethanthedistanceofthesamelayer'sadjacentpixel.IftheseCTimagedataaredirectlyusedby3Ddisplaying,theboundaryofreconstructiveobjectswillsignificantlyappearstepphenomenon,CTimagemultiplecontourreconstructionisthekeytosolvetheproblem.Itextractseverylayerimageedgecontourbyedgedetectionalgorithm,convertsthemultiplecontourconnectiontothesinglecontourconnectionusingauxiliarycontours.Itreconstructsobjectbystructuringacceptablesurface,finallyobtains3Dsurfacereconstructionobjectsandimprovesthe3Dsurfacereconstructiveobject’squality.Accordingtothemedicalimage'scharacteristicsofgrayfuzzinessandinhomogeneity,thepapercenteringonthetechnologytheoryofmedicalimage3Dsurfacereconstructionemphasislydiscussestheedgedetectionandmultiplecontourreconstruction.Researchingexistingedgedetectionand3Dreconstructionandcomparingthem,animprovededgedetectionalgorithmandauxiliarycontoursreconstructionisproposedaimingtoexistingproblemofedgedetectionand3Dsurfacereconstruction,themainworkcanbesummarizedsomebelowaspects:Firstly,itproposesamethodofmultiplescaleandanglestructuralelementedgedetection.Existingedgedetectionoperatorcommonlyexiststheproblemofineffectivelysolvinglocatingedgeandrestrainingnoise,thealgorithmimprovesinstructuringmultiplescaleelementandfuzzymorphologicaloperation.Structuringmultiplescaleandanglestructuralelementandfuzzymorphologicaloperationisthemainoperatedinthemethod.Itbasesontheimagequad-treedecompositiontechnologyandfuzzymorphologicaltechnology,comprehensivelyconsideringthedifferentofimagepixelandavoidsstructuringcomplexstructuralelementforIVincreasingthecomplexofalgorithm.Inaddition,consideringthemedicalimagecharacterofgrayfuzzinessandinhomogeneity,itoperatesimagewithfuzzymorphologicalinordertoshowclearlytheimageboundarycontour,welldistinguishingtheedgeandnoise.Itisprovedthatthemethodiseffectonthetheoryanalysisandexperimentalverificationandobtainsthebetterexperimentalresults.Secondly,itproposesanalgorithmofimprovedmultiplecontour3Dsurfacereconstruction.Aimingthatthecurrentstatusofexisting3Dsurfacereconstructionalgorithmexiststheproblemoflargecomputationalcomplexityandlongconsumingtime,thealgorithmimprovesinstructuringauxiliarycontours;itextractsimage'sedgeusingfuzzymorphological;itstructuresthecharacterpointofeverycontourusingthemethodofcontourapproximation.Itrealizesthatthethreekeratinizationofcharacteristicpointusingtheacceptablesurface.Inthecourseofreconstruction,consideringthenotsmoothphenomenon,itoperatesthemethodofthemostlesttwosquareapproximationsonthepointofcontoursurface,realizingCTmedicalimage's3Dsurfacereconstruction.Itisprovedintheexperimentthatusingthealgorithmshortenthetimeofvolumedatastructuretimeatthesametimeofensurethecorrectionofreconstructionobjects,acceleratingthespeedofthewholesurfacereconstruction.Finally,theabovealgorithmisrealizedusingMATLABprogram,verifytheabovealgorithmbyexperimentdataandcompareexistingalgorithm.Itisprovedintheexperimentthattheedgedetectionalgorithmdesignedbythepaperimprovesthelackofpreviousalgorithm,extractededgecontourcanwellshowinnerobject'sdetail,solvesthecontradictionbetweennoiseandedgepoint.Thereconstructedobjectcanwellshowobject's3Dsurfaceinformationbasedonextractedcontours.Theruntimeislessthanelsealgorithm,thealgorithmsolvestheproblemofcurrentalgorithmlongruntimeandlowefficiency.KEYWORDS:edgedetection,fuzzymorphological,structuralelementmultiplecontour,3DsurfacereconstructionsDissertationType:ResearchonApplication48CTComputerizedTomographyMRIMagneticResonanceImaging2DTwoDimensional3DThreeDimensionalRLERunLengthEncoded1111.1(VisualizationinScientificComputing)[1,2]2080(CTComputedTomography)(MRIMagneticResonanceImaging)(USUltraSonogram)(MEG)(IR)[3][4][5]CTMRI2[6]1.21.2.1[7][8]1.2.21.()13[9]()[10](1)[11]RobertsPrewittSobelKirsch[12](2)Hueckell[13]Haralick[14]Haralick(3)Marr-Hildreth[15]LOG(LaplacianofGaussian)4LOGLOGCannyCanny[16]CannyCannyLOGCanny(4)Rosenfeld[17]Marr[18]Witkin[19]Canny[16][20][21][22]Har(5)PalKing[23]15(6)Soinel[24](7)(8)FesharakiHellestrand[25]555HouKoh[26]2.6(SurfaceFiting),(DirectVolumeRendering)(HybridApproaches)(1)CT(ContourConnection)(ShapefromPlanarContours)Keppel1975[27]CTMRI()(Cuberille)(MarchingCubes,MC)(DividingCubes)Lorensen1987(MC)[28]MCM.J.DurstMC[29]LorensenMC(MarchingTetrahedron)[30]MCMCMC17(2)CT.MRICT(Image-order)(Object-order)(Hybrid-order)(RayTracingAlgorithm)(BackwardProjectingAlgorithm)(slatingAlgorithm)(FootingAlgorithm)[31](ForwardProjectingAlgorithm)(Shear-WarpFactorizationAlgorithm)8(DCT)()(FFT&FFT)(3)Levoy[32]GoodsellWalsunl[