KTH-CSC / CVAP / Courses / Multiple View Geometry 2010 / Description


Multiple View Geometry Course


Description

A basic problem in computer vision is to reconstruct a real world scene given several images of it. Multiple View Geometry course presents techniques for solving this problem that are taken from projective geometry and photogrammetry. It is organized as a reading course based on the R. Hartley and A. Zisserman's book Multiple View Geometry in Computer Vision (see also Course Materials), and is an extension of DD2428 Geometric Computing and Visualization (Computational Photography). The course is given at the PhD level, and as such deals with more advanced topics and problems than DD2428 and requires a larger amount of self study.
Meetings

The course includes nine two-hour meetings which are scheduled every second Thursday. The first meeting will include an introductory presentation given by a tutor, followed by presentation of Chapter 2 of the course book prepared by students (see Schedule & Activity). The meetings 2-9 will consist of the two parts:
  • (1) an exercise session supervised by a tutor where problems related to the previous meeting will be discussed;
  • (2) a presentation of a new chapter of the course book prepared by pre-assigned students (see Schedule & Activity).
All students should read a relevant part of the book before the meeting in which it will be presented to be able to take active part in discussion and ask questions in case of any doubts. Homework assignments will be handed out at the meetings. The deadline for handing in solutions to the assignment handed out at the meeting N is at the meeting N+1 (read more about requirements).

If you are not taking the course you are still very welcome to join the presentation part of the meetings (every second Thursday 3-4:30pm, see Schedule & Activity).
Excerpt from course plan

Goals
  • To understand the geometric relations between multiple views of scenes.
  • To understand the general principles of parameter estimation.
  • To be able to compute scene and camera properties from real world images using state-of-the-art algorithms.

Syllabus
  • PART 0 - Projective Geometry, Transformation and Estimation: Projective geometry and transformations of 2D and 3D; Estimation of 2D projective transformation.
  • PART 1 - Camera Geometry and Single View Geometry: camera models; computation of the camera matrix; action of a projective camera on planes, lines, conics and quadrics.
  • PART 2 - Two-View Geometry: epipolar geometry and the fundamental matrix; 3D reconstruction of cameras and structure; different algorithms for computing the fundamental matrix.
  • PART 3 - N-View Geometry: N-view computational methods; auto-calibration.

Prerequisites
  • Knowledge corresponding to the compulsory courses on mathematics, computer science and numerical analysis on D-, E- or F-programme.
  • Recommended: DD2428 Geometric Computing and Visualization (Computational Photography) or equivalent.

Course book
Richard Hartley and Andrew Zisserman Multiple View Geometry in Computer Vision. Second Edition, Cambridge University Press, March 2004.

See also Course Materials.



[Course homepage]