Epipolar Geometryclass 11.ppt

Epipolar Geometryclass 11 Multiple View Geometry Comp 290-089 Marc Pollefeys Content Background: Projective geometry (2D, 3D), Parameter estimation, Algorithm evaluation. Single View: Camera model, Calibration, Single View Geometry. Two Views: Epipolar Geometry, 3D reconstruction, Computing F, Computing structure, Plane and homographies. Three Views: Trifocal Tensor, Computing T. More Views: N-Linearities, Multiple view reconstruction, Bundle adjustment, auto-calibration, Dynamic SfM, Cheirality, Duality Multiple View Geometry course schedule(subject to change) More Single-View Geometry Projective cameras and planes, lines, conics and quadrics. Camera calibration and vanishing points, calibrating conic and the IAC Next class: 3D reconstruction Project Demos Duality Apr. 22, 24 Papers Cheirality Apr. 15, 17 Papers Dynamic SfM Apr. 8, 10 Papers Auto-Calibration Apr. 1, 3 Bundle adjustment MultipleView Reconstruction Mar. 25, 27 Multiple View Geometry Three View Reconstruction Mar. 18, 20 Trifocal Tensor Planes Homographies Mar. 4, 6 Structure Comp. Fund. Matrix Comp. Feb. 25, 27 3D reconstruction Epipolar Geometry Feb. 18, 20 Single View Geometry Camera Calibration Feb. 11, 13 Camera Models Algorithm Evaluation Feb. 4, 6 Parameter Estimation Parameter Estimation Jan. 28, 30 (no class) Projective 3D Geometry Jan. 21, 23 Projective 2D Geometry (no class) Jan. 14, 16 Projective 2D Geometry Intro motivation Jan. 7, 9 Two-view geometry Epipolar geometry 3D reconstruction F-matrix comp. Structure comp. Correspondence geometry: Given an image point x in the first view, how does this constrain the position of the corresponding point x’ in the second image? (ii) Camera geometry (motion): Given a set of corresponding image points {xi ?x’i}, i=1,…,n, what are the cameras P and P’ for the two views? (iii) Scene geometry (structure): Given corresponding image points xi ?x’i and cameras P, P’, what is the position of (their pre-image) X in space? Three ques