3D Urban ModelingSlide 23D Urban Modeling course schedule (tentative)Course materialGeometry primerSlide 6Geometry primer Hierarchy of 2D transformationsGeometry primer Hierarchy of 3D transformationsCamera modelProjector modelSlide 11Slide 12Slide 13Slide 14Slide 15Slide 17Slide 18Slide 19Slide 20Slide 21Slide 22Radial distortionSlide 24Meydenbauer cameraSlide 26Slide 27Slide 28Slide 29Slide 30Singular Value DecompositionSlide 32Slide 33Slide 34Slide 35Gold Standard algorithmSlide 37Slide 38Slide 39Slide 40Slide 41Image of absolute conicSlide 43Slide 44Self-calibration using absolute conicPractical linear self-calibration3D Urban ModelingMarc PollefeysJan-Michael Frahm, Philippos MordohaiFall 2006 / Comp 790-089Tue & Thu 14:00-15:153D Urban Modeling•Basics: sensor, techniques, …•Video, LIDAR, GPS, IMU, …•SfM, Stereo, …•Robot Mapping, GIS, …•Review state-of-the-art•Video and LIDAR-based systems•Projects to experiment•Virtual UNC-Chapel Hill•DARPA Urban Challenge•…3D Urban Modeling course schedule(tentative)Sep. 5, 7 IntroductionVideo-based Urban 3D CaptureSep. 12, 14 CamerasSep. 19, 21Sep. 26, 28 Project proposalsOct. 3, 5Oct. 10, 12Oct. 17, 19 (fall break)Oct. 24, 26 Project Status UpdateOct. 31, Nov. 2Nov. 7, 9Nov. 14, 16Nov. 21, 23 (Thanksgiving)Nov. 28, 30Dec. 5 Project demonstrations (classes ended)Note: Dec. 3 is CVPR deadlineCourse materialSlides, notes, papers and referencessee course webpage/wiki (later)On-line “shape-from-video” tutorial:http://www.cs.unc.edu/~marc/tutorial.pdfhttp://www.cs.unc.edu/~marc/tutorial/Geometry primerl'lx Incidence of point and line x'xl 0xl T0Xπ T0ll*CT1* CC0QXX T0πQπ*T-1*QQ JoinsConics and dual conics0xx CTIncidence of point and plane JoinsConics and dual conics0XπππT3T2T10πXXXT3T2T1Geometry primerTransformation for linesll'-THTransformation for conics-1-TCHHC 'Transformation for dual conicsTHHCC**' xx' HTransformation for pointsXX' Hππ'-TH-1-TQHHQ'THHQQ'**Transformation for planesTransformation for quadricsTransformation for dual quadricsTransformation for pointsGeometry primer Hierarchy of 2D transformations10022211211yxtaataa10022211211yxtsrsrtsrsr333231232221131211hhhhhhhhh10022211211yxtrrtrrProjective8dofAffine6dofSimilarity4dofEuclidean3dofConcurrency, collinearity, order of contact (intersection, tangency, inflection, etc.), cross ratioParallellism, ratio of areas, ratio of lengths on parallel lines (e.g midpoints), linear combinations of vectors (centroids). The line at infinity l∞Ratios of lengths, angles.The circular points I,Jlengths, areas.invariantstransformed squaresGeometry primer Hierarchy of 3D transformationsvTvtAProjective15dofAffine12dofSimilarity7dofEuclidean6dofIntersection and tangencyParallellism of planes,Volume ratios, centroids,The plane at infinity π∞Angles, ratios of lengthThe absolute conic Ω∞Volume10tAT10tRTs10tRTCamera modelRelation between pixels and rays in space?Projector modelRelation between pixels and rays in space(dual of camera)(main geometric difference is vertical principal point offset to reduce keystone effect)?Affine camerasPinhole camera modelTTZfYZfXZYX )/,/(),,( 101001ZYXffZfYfXZYXlinear projection in homogeneous coordinates!homogeneous coordinatesnon-homogeneous coordinatesPinhole camera model10100ZYXffZfYfX10101011ZYXffZfYfXPXx 0|I)1,,(diagP ffPrincipal point offsetTyxTpZfYpZfXZYX )/,/(),,( principal pointTyxpp ),(101001ZYXpfpfZZpfYZpfXZYXyxxxPrincipal point offset10100ZYXpfpfZZpfYZpfXyxxx camX0|IKx 1yxpfpfKcalibration matrixObject motionCamera motionCCD camera1yyxxppK11yxyxpfpfmmKGeneral projective camera1yxxxppsK1yxxxppK t|IKRP non-singular11 dof (5+3+3) t|RKP intrinsic camera parametersextrinsic camera parametersCamera matrix decompositionFinding the camera center0PC (use SVD to find null-space)Finding the camera orientation and internal parametersKR(use RQ decomposition ~QR)QR=( )-1= -1 -1QR(if only QR, invert)PXλC)P(Xx (for all X and λ C must be camera center) t|RKP Affine camerasRadial distortion•Due to spherical lenses (cheap)•Model:RyxyxKyxKyx ...))()(1(),(2222221Rhttp://foto.hut.fi/opetus/260/luennot/11/atkinson_6-11_radial_distortion_zoom_lenses.jpgstraight lines are not straight anymorepincushion dist.barrel dist.Radial distortion exampleMeydenbauer cameravertical lens shiftto allow direct ortho-photographsAction of projective camera on points and linesforward projection of line μbaμPBPAμB)P(AμX back-projection of linelPTPXlXTT PX x0;xlTPXx projection of pointAction of projective camera on conics and quadricsback-projection to coneCPPQTco0CPXPXCxxTTT PXx projection of quadricTPPQC**0lPPQlQT*T*T lPTiixX ? PResectioningDirect Linear Transform (DLT)iiPXx iiPXxrank-2 matrixDirect Linear Transform (DLT)Minimal solutionOver-determined solution 5½ correspondences needed (say 6) P has 11 dof, 2 independent eq./pointsn 6 pointsApminimize subject to constraint 1p use SVDSingular Value DecompositionXXVTXVΣTXVUΣTHomogeneous least-squaresTVUΣA 1X AXminsubject tonVX solutionTVΣUXDegenerate configurations(i) Points lie on
View Full Document