3D VisionStereo VisionEpipolar GeometrySlide 4Slide 5Essential MatrixSlide 7Fundamental MatrixSlide 9Computing F: The Eight-point AlgorithmLocating the Epipoles from FStereo RectificationSlide 13Slide 14Slide 15Slide 16Correspondence problemCorrelation ApproachSlide 19Slide 20Slide 21Slide 22Slide 23Slide 24Feature-based ApproachSlide 26Slide 27Slide 28Advanced TopicsSlide 30Slide 31Slide 323D Reconstruction ProblemReconstruction by TriangulationReconstruction up to a Scale FactorReconstruction up to a Projective TransformationSummaryNext3D Computer Visionand Video Computing3D Vision3D Vision Topic 7 of Part 2 Stereo Vision (II)CSC I6716Spring 2004Zhigang Zhu, NAC 8/203Ahttp://www-cs.engr.ccny.cuny.edu/~zhu/VisionCourse-2004.html3D Computer Visionand Video ComputingStereo VisionStereo VisionEpipolar GeometryWhere to search correspondencesEpipolar plane, epipolar lines and epipolesEssential matrix and fundamental matrixCorrespondence ProblemCorrelation-based approachFeature-based approach 3D Reconstruction ProblemBoth intrinsic and extrinsic parameters are knownOnly intrinsic parametersNo prior knowledge of the cameras3D Computer Visionand Video ComputingEpipolar GeometryEpipolar GeometryNotationsPl =(Xl, Yl, Zl), Pr =(Xr, Yr, Zr) Vectors of the same 3-D point P, in the left and right camera coordinate systems respectivelyExtrinsic ParametersTranslation Vector T = (Or-Ol) Rotation Matrix Rpl =(xl, yl, zl), pr =(xr, yr, zr)Projections of P on the left and right image plane respectivelyFor all image points, we have zl=fl, zr=frT)R(PPlrlPplllZfrrrrZfPp plprPOlOrXlXrPlPrflfrZlYlZrYrR, T3D Computer Visionand Video ComputingEpipolar GeometryEpipolar GeometryMotivation: where to search correspondences?Epipolar PlaneA plane going through point P and the centers of projection (COPs) of the two camerasConjugated Epipolar Lines Lines where epipolar plane intersects the image planesEpipolesThe image in one camera of the COP of the otherEpipolar ConstraintCorresponding points must lie on conjugated epipolar linesplprPOlOrelerPlPrEpipolar PlaneEpipolar LinesEpipoles3D Computer Visionand Video ComputingEpipolar GeometryEpipolar GeometryMotivation: where to search correspondences?Epipolar PlaneA plane going through point P and the centers of projection (COPs) of the two camerasConjugated Epipolar Lines Lines where epipolar plane intersects the image planesEpipolesThe image in one camera of the COP of the otherEpipolar ConstraintCorresponding points must lie on conjugated epipolar linesplprPOlOrelerPlPrEpipolar PlaneEpipolar LinesEpipoles3D Computer Visionand Video ComputingEssential MatrixEssential MatrixEquation of the epipolar planeCo-planarity condition of vectors Pl, T and Pl-TEssential Matrix E = RS 3x3 matrix constructed from R and T (extrinsic only)Rank (E) = 2, two equal nonzero singular values0llPTT)(PT000xyxzyzTTTTTTS333231232221131211rrrrrrrrrRRank (R) =3Rank (S) =2T)R(PPlr0lTrEPP0lTrEpplPplllZfrrrrZfPp 3D Computer Visionand Video ComputingEssential MatrixEssential MatrixEssential Matrix E = RS A natural link between the stereo point pair and the extrinsic parameters of the stereo system One correspondence -> a linear equation of 9 entriesGiven 8 pairs of (pl, pr) -> EMapping between points and epipolar lines we are looking forGiven pl, E -> pr on the projective line in the right planeEquation represents the epipolar line of either pr (or pl) in the right (or left) image Note: pl, pr are in the camera coordinate system, not pixel coordinates that we can measure0lTrEpp3D Computer Visionand Video ComputingFundamental MatrixFundamental MatrixMapping between points and epipolar lines in the pixel coordinate systemsWith no prior knowledge on the stereo systemFrom Camera to Pixels: Matrices of intrinsic parametersQuestions: What are fx, fy, ox, oy ?How to measure pl in images?0lTrpFp1lrEMMFTl1llpMprrrpMp110000int yyxxofofM0lTrEppRank (Mint) =33D Computer Visionand Video ComputingFundamental MatrixFundamental MatrixFundamental Matrix Rank (F) = 2Encodes info on both intrinsic and extrinsic parametersEnables full reconstruction of the epipolar geometryIn pixel coordinate systems without any knowledge of the intrinsic and extrinsic parameters Linear equation of the 9 entries of F0lTrpFp1lrEMMFT01333231232221131211)1()()()()(rimrimlimlimyxfffffffffyx3D Computer Visionand Video ComputingComputing F: The Eight-point AlgorithmComputing F: The Eight-point AlgorithmInput: n point correspondences ( n >= 8)Construct homogeneous system Ax= 0 fromx = (f11,f12, ,f13, f21,f22,f23 f31,f32, f33) : entries in FEach correspondence give one equationA is a nx9 matrixObtain estimate F^ by SVD of Ax (up to a scale) is column of V corresponding to the least singular valueEnforce singularity constraint: since Rank (F) = 2Compute SVD of F^Set the smallest singular value to 0: D -> D’Correct estimate of F : Output: the estimate of the fundamental matrix, F’Similarly we can compute E given intrinsic parameters0lTrpFpTUDVA TUDVF ˆTVUDF' '3D Computer Visionand Video ComputingLocating the Epipoles from FLocating the Epipoles from FInput: Fundamental Matrix FFind the SVD of FThe epipole el is the column of V corresponding to the null singular value (as shown above)The epipole er is the column of U corresponding to the null singular value (similar treatment as for el)Output: Epipole el and erTUDVF el lies on all the epipolar lines of the left image0lTrpFp0lTreFpF is not identically zeroFor every pr0leFplprPOlOrelerPlPrEpipolar PlaneEpipolar LinesEpipoles3D Computer Visionand Video ComputingStereo RectificationStereo RectificationRectification Given a stereo pair, the intrinsic and extrinsic parameters, find the image transformation to achieve a stereo system of horizontal epipolar linesA simple algorithm: Assuming calibrated stereo camerasp’lp’rPOlOrX’rPlPrZ’lY’lY’rTX’lZ’rStereo System with Parallel Optical AxesEpipoles are at
View Full Document