SegmentationandBoundaryDetectionUsingMultiscale利用多尺分割和邊界檢測_第1頁
SegmentationandBoundaryDetectionUsingMultiscale利用多尺分割和邊界檢測_第2頁
SegmentationandBoundaryDetectionUsingMultiscale利用多尺分割和邊界檢測_第3頁
SegmentationandBoundaryDetectionUsingMultiscale利用多尺分割和邊界檢測_第4頁
SegmentationandBoundaryDetectionUsingMultiscale利用多尺分割和邊界檢測_第5頁
已閱讀5頁，還剩59頁未讀， 繼續(xù)免費(fèi)閱讀

下載本文檔

版權(quán)說明：本文檔由用戶提供并上傳，收益歸屬內(nèi)容提供方，若內(nèi)容存在侵權(quán)，請(qǐng)進(jìn)行舉報(bào)或認(rèn)領(lǐng)

文檔簡介

SegmentationandBoundaryDetectionUsingMultiscale利用多尺分割和邊界檢測第1頁/共64頁LocalUncertainty第2頁/共64頁GlobalCertainty第3頁/共64頁LocalUncertainty第4頁/共64頁GlobalCertainty第5頁/共64頁CoarseMeasurementsforTexture第6頁/共64頁AChickenandEggProblemProblem:CoarsemeasurementsmixneighboringstatisticsSolution:supportofmeasurementsisdeterminedasthesegmentationprocessproceeds第7頁/共64頁

Normalized-cutsmeasureingraphsCompletehierarchyinlineartimeUsemultiscalemeasuresofintensity,texture,shape,andboundaryintegritySegmentationbyWeightedAggregation第8頁/共64頁Normalized-cutsmeasureingraphs

CompletehierarchyinlineartimeUsemultiscalemeasuresofintensity,texture,shape,andboundaryintegritySegmentationbyWeightedAggregation第9頁/共64頁SegmentationbyWeightedAggregationNormalized-cutsmeasureingraphsCompletehierarchyinlineartime

Usemultiscalemeasuresofintensity,texture,shapeandboundaryintegrity第10頁/共64頁ThePixelGraphCouplingsReflectintensitysimilarityLowcontrast–strongcouplingHighcontrast–weakcoupling第11頁/共64頁HierarchicalGraph第12頁/共64頁Hierarchy

inSWA第13頁/共64頁Normalized-CutMeasure第14頁/共64頁Normalized-CutMeasure第15頁/共64頁Normalized-CutMeasure第16頁/共64頁Normalized-CutMeasureMinimize:第17頁/共64頁Normalized-CutMeasureHigh-energycutMinimize:第18頁/共64頁Normalized-CutMeasureLow-energycutMinimize:第19頁/共64頁RecursiveCoarsening第20頁/共64頁RecursiveCoarseningRepresentativesubset第21頁/共64頁RecursiveCoarseningForasalientsegment:,sparseinterpolationmatrix第22頁/共64頁WeightedAggregationaggregateaggregate第23頁/共64頁SegmentDetection第24頁/共64頁SWALinearin#ofpoints(afewdozenoperationsperpoint)DetectsthesalientsegmentsHierarchicalstructure第25頁/共64頁Coarse-ScaleMeasurementsAverageintensitiesofaggregatesMultiscaleintensity-variancesofaggregatesMultiscaleshape-momentsofaggregatesBoundaryalignmentbetweenaggregates第26頁/共64頁Adaptivevs.RigidMeasurementsAveragingOuralgorithm-SWAGeometricOriginal第27頁/共64頁Ouralgorithm-SWAAdaptivevs.RigidMeasurementsInterpolationGeometricOriginal第28頁/共64頁RecursiveMeasurements:Intensityaggregateintensityofpixeli

averageintensityofaggregate第29頁/共64頁UseAveragestoModifytheGraph第30頁/共64頁UseAveragestoModifytheGraph第31頁/共64頁TextureExamples第32頁/共64頁IsotropicandOrientedFiltersTextonsbyK-MeansMaliketalIJCV2001Abrieftutorial第33頁/共64頁IsotropicTextureinSWAIntensityVarianceIsotropicTextureofaggregate–averageofvariancesinallscales第34頁/共64頁IsotropicTextureinSWAIntensityVarianceIsotropicTextureofaggregate–averageofvariancesinallscales第35頁/共64頁IsotropicTextureinSWAIntensityVarianceIsotropicTextureofaggregate–averageofvariancesinallscales第36頁/共64頁OrientedTextureinSWAShapeMomentsOrientedTextureofaggregate–orientation,widthandlengthinallscalescenterofmasswidthlengthorientationwithMeiravGalun第37頁/共64頁Gestalt–PerceptualGroupingAbriefTutorialSharon,Brandt,BasriPAMI2000:ShashuaandUllmanICCV1988:Groupcurvesby:ProximityCo-linearity第38頁/共64頁BoundaryIntegrityinSWA第39頁/共64頁SharpentheAggregatesTop-downSharpening:

ExpandcoreSharpenboundaries第40頁/共64頁Hierarchy

inSWA第41頁/共64頁ExperimentsOurSWAalgorithm(CVPR’00+CVPR’01)

run-time:5-10seconds.Normalizedcuts(ShiandMalik,PAMI’00;Maliketal.,IJCV’01)

run-time:about10-15minutes.

SoftwarecourtesyofDoronTal,UCBerkeley.imagesonapentiumIII1000MHzPC:第42頁/共64頁IsotropicTexture-HorseIOurAlgorithm(SWA)NormalizedCuts第43頁/共64頁IsotropicTexture-HorseIIOurAlgorithm(SWA)NormalizedCuts第44頁/共64頁IsotropicTexture-TigerNormalizedCutsOurAlgorithm(SWA)第45頁/共64頁IsotropicTexture-ButterflyOurAlgorithm(SWA)NormalizedCuts第46頁/共64頁IsotropicTexture-LeopardOurAlgorithm(SWA)第47頁/共64頁IsotropicTexture-DalmatianDogOurAlgorithm(SWA)第48頁/共64頁IsotropicTexture-SquirrelOurAlgorithm(SWA)NormalizedCuts第49頁/共64頁FullTexture-SquirrelOurAlgorithm(SWA)NormalizedCutswithMeiravGalun第50頁/共64頁FullTexture-CompositionOurAlgorithm(SWA)withMeiravGalun第51頁/共64頁FullTexture–LionCubOurAlgorithm(SWA)withMeiravGalun第52頁/共64頁FullTexture–PolarBearOurAlgorithm(SWA)withMeiravGalun第53頁/共64頁FullTexture–PenguinOurAlgorithm(SWA)withMeiravGalun第54頁/共64頁FullTexture–LeopardOurAlgorithm(SWA)withMeiravGalun第55頁/共64頁FullTexture-ZebraOurAlgorithm(SWA)withMeiravGalun第56頁/共64頁SegmentationbyWeightedAggregationEfficientapproximationtoNcut-likemeasuresRecursivecomputationofmultiscalemeasurementsNoveladaptivepyramidrepresentingtheimage第57頁/共64頁MatchingExperiment(ChenBrestel)第58頁/共64頁MatchingExperiment(ChenBrestel)第59頁/共64頁MatchingExperiment(ChenBrestel)第60頁/共64頁Experiments–ClusteringSilho

溫馨提示

  • 1. 本站所有資源如無特殊說明，都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請(qǐng)下載最新的WinRAR軟件解壓。
  • 2. 本站的文檔不包含任何第三方提供的附件圖紙等，如果需要附件，請(qǐng)聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
  • 3. 本站RAR壓縮包中若帶圖紙，網(wǎng)頁內(nèi)容里面會(huì)有圖紙預(yù)覽，若沒有圖紙預(yù)覽就沒有圖紙。
  • 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
  • 5. 人人文庫網(wǎng)僅提供信息存儲(chǔ)空間，僅對(duì)用戶上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理，對(duì)用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯，并不能對(duì)任何下載內(nèi)容負(fù)責(zé)。
  • 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容，請(qǐng)與我們聯(lián)系，我們立即糾正。
  • 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時(shí)也不承擔(dān)用戶因使用這些下載資源對(duì)自己和他人造成任何形式的傷害或損失。

最新文檔

評(píng)論

0/150

提交評(píng)論