TeachingSs07CV2

Chair for Computer Aided Medical Procedures & Augmented Reality
Lehrstuhl für Informatikanwendungen in der Medizin & Augmented Reality

3D Computer Vision II Summer Term 2007 - Thu, Jul 19, 2007 ALERT! Final exam July 19 at 10:30 in MI 03.13.008

Administrative Info

Lecture by Prof. Nassir Navab
Exercises by Dr. Selim Benhimane and Hauke Heibel

2+2 SWS, 5 ECTS, Theoretische Informatik, Wahlpflichtfach

Time, Location & Requirements

Tuesday 10:30-12:00 MI 03.13.010
Exercises: Thursday 13:00-14:30 MI 03.13.008

Requirements:

* The classes and exam are in English
* The final exam is written and only notes are allowed (no book)
* The final exam contains 100 points, you need to have 50 points to pass it
* Up to 40 bonus points can be earned from the homework and the intermediate exam wink

Site Content

Announcements

First Lecture: Tuesday, 17 April 2007 at 10:30-12:00 MI 03.13.010
First Exercises: Thursday, 19 April 2007 at 13:00-14:30 MI 03.13.010

Homework must be handed over at latest until the beginning of the following exercise. Programming assignments should be sent to the person who conducted the exercise in which the homework was given. Written assignments should be put in the blue homework box at the chair (in front of room MI 03.13.052).

Solutions for the exercises and homework can be found on the web-site one week after the exercise in which the homework was given. Final grades will only be given at the end of the term. If somebody needs personal feedback, you can drop by in the office of the person who conducted the exercise in which the homework was given.

ALERT! Final Exam in MI 03.13.010 (90 minutes) ALERT! at 10:30
No book, no lecture slides allowed.
Only a two-sided written A4 paper is allowed during the test .
It contains your own notes and should be hand written.

Overview

Making a computer see was something that leading experts in the field of Artificial Intelligence thought to be at the level of difficulty of a summer student's project back in the sixties. Forty years later the task is still unsolved and seems formidable. A whole field, called Computer Vision, has emerged as a discipline in itself with strong connections to mathematics and computer science and looser connections to physics, the psychology of perception and the neuro sciences.

Over the past decade there has been a rapid development in the understanding and modelling of the geometry of multiple views in computer vision. The theory and practice have now reached a level of maturity where excellent results can be achieved for problems that were unsolved a decade ago, and often thought unsolvable. These tasks and algorithms include problems like:
Given two/three/multiple images, and no further information, compute/estimate:

  • matches between the images
  • the 3D position of the points that generate these matches
  • the cameras that generate the images

adapted form Hartley & Zisserman's "Multiple View Geometry in Computer Vision"

This tasks and algorithms as well as the methods which allow us to reason about the quality of our results are the core of the lecture 3D Computer Vision II. This lecture deals with multiple view geometry problems and having attended 3D Computer Vision I is of great benefit, however, it is not, in any case, compulsory.

Content

  1. Intro, motivation & overview
  2. Projective geometry and 2D, 3D transformations
  3. Parameter estimation
  4. Camera model
  5. Two-View: Epipolar Geometry, 3D reconstruction, Fundamental Matrix, Structure computation
  6. Two-View: Mixed Camera Models: Perspective and Orthographic, Combined Othrographic and Perspective (COP) Images
  7. Three-View: The Trifocal Tensor
  8. Multi-View: Factorization Method, Multi-view Reconstruction, Motion and Structure from Motion

Course Schedule

Date Type Topic Conducted by Material/Literature
Tue, Apr 17, 2007 C Introduction and Overview Prof. Dr. N. Navab slides1.pdf
Tue, Apr 24, 2007 C Projective Geometry Prof. Dr. N. Navab 3d_CV_Lecture01.pdf
Tue, May 1, 2007   no class due to May Day      
Tue, May 8, 2007 C Parameter estimation Prof. Dr. N. Navab 3d_CV_Lecture02.pdf  
Tue, May 15, 2007   no class due to SVV      
Tue, May 22, 2007 C One-view Geometry (part I) Prof. Dr. N. Navab 3d_CV_Lecture03.pdf  
Tue, May 29, 2007   no class due to Pentecost      
Tue, Jun 5, 2007 C Two-View Geometry: Epipolar Geometry Dr. S. Benhimane    
Tue, Jun 12, 2007 C Two-View Geometry: 3D reconstruction Prof. Dr. N. Navab 3d_CV_Lecture04.pdf  
Tue, Jun 19, 2007 C Two-View Geometry: Fundamental Matrix Prof. Dr. N. Navab 3d_CV_Lecture05.pdf  
Tue, Jun 26, 2007 C Two-View Geometry: Structure computation Prof. Dr. N. Navab 3d_CV_Lecture06_01.pdf
and
3d_CV_Lecture06_02.pdf
 
Tue, Jul 3, 2007 I Invited Talk: Dr. Mirko Appel (Siemens Corporate Technology) Dr. Mirko Appel    
Tue, Jul 10, 2007 C Three-View Geometry Prof. Dr. N. Navab 3d_CV_TriFocalTensor.pdf  
Tue, Jul 17, 2007 E ALERT! Multiple-View Geometry (Exercise ALERT!) Dr. S. Benhimane    

Exercises

Date Type Topic Conducted by Material/Literature Assignment/Solution
Thu, Apr 19, 2007 E Optional if you did 3D Computer Vision I (WS 2006-2007)
Introduction to numerical computing with Matlab
Homogeneous Representation of Points and Lines in Projective Space
Vector Notation
Simple Coordinate Transformations
Dr. Selim Benhimane    
Thu, Apr 26, 2007 E Optional if you did 3D Computer Vision I (WS 2006-2007)
Projective Geometry in 2D
Cross-Ratio of Angles
The Cross-Ratio in Needle Placement
The Cross-Ratio for Depth-Recovery
Dr. Selim Benhimane    
Thu, May 3, 2007 E Hierarchy of Transformations
Invariants under Transformations
Rotations in 3D
Hauke Heibel    
Thu, May 10, 2007 E SVD
DLT
Pseudo-Inverse
Algebraic vs. Geometric Error
Hauke Heibel    
Thu, May 17, 2007   no class due to Ascension Day      
Thu, May 24, 2007 E Projection Matrices Hauke Heibel    
Thu, May 31, 2007 C One-view Geometry (part II) Prof. Dr. N. Navab    
Thu, Jun 7, 2007   no class due to Feast of Corpus Christi      
Thu, Jun 14, 2007 T Intermediate Exam in MI 03.13.010 (90 minutes)
No book, no lecture slides allowed.
Only a two-sided written A4 paper is allowed during the test .
It contains your own notes and should be hand written.
Dr. Selim Benhimane    
Thu, Jun 21, 2007 E Metric Reconstruction Hauke Heibel    
Thu, Jun 28, 2007 E The optimal triangulation method Dr. Selim Benhimane    
Thu, Jul 5, 2007 E Homography and Fundamental matrix linear computations Dr. Selim Benhimane    
Thu, Jul 12, 2007 C Multiple-View Geometry (Lecture) Dr. Selim Benhimane    
Thu, Jul 19, 2007ALERT! at 10:30
T Final Exam in MI 03.13.010 (90 minutes) ALERT! at 10:30
No book, no lecture slides allowed.
Only a two-sided written A4 paper is allowed during the test .
It contains your own notes and should be hand written.
Dr. Selim Benhimane    

Readings

  • Primary Reading
    • Multiple View Geometry in Computer Vision by Richard Hartley & Andrew Zisserman

  • General Introduction to 3D Computer Vision
    • Three-Dimensional Computer Vision by Olivier Faugeras
    • Computer Vision: A Modern Approach by David A. Forsyth & Jean Ponce
    • Introductory Techniques for 3-D Computer Vision by Emanuele Trucco & Alessandro Verri

  • More Specific Readings
    • The Geometry of Multiple Images: The Laws That Govern the Formation of Multiple Images of a Scene and Some of Their Applications by Olivier Faugeras, Quang-Tuan Luong, Theodore H. Papadopoullos; MIT Press; 2001


TeachingForm
Title: 3D Computer Vision II
Professor: Prof. Dr. Nassir Navab
Tutors: Dr. Selim Benhimane and Hauke Heibel
Type: Lecture
Information: 2+2 SWS, 5 ECTS, Theoretische Informatik, Vertiefungsfach
Term: 2007SoSe
Abstract: Making a computer see was something that leading experts in the field of Artificial Intelligence thought to be at the level of difficulty of a summer student's project back in the sixties. Forty years later the task is still unsolved and seems formidable. A whole field, called Computer Vision, has emerged as a discipline in itself with strong connections to mathematics and computer science and looser connections to physics, the psychology of perception and the neuro sciences. Over the past decade there has been a rapid development in the understanding and modelling of the geometry of multiple views in computer vision. The theory and practice have now reached a level of maturity where excellent results can be achieved for problems that were unsolved a decade ago, and often thought unsolvable. These tasks and algorithms include problems like: Given two/three/multiple images, and no further information, compute/estimate: * matches between the images * the 3D position of the points that generate these matches * the cameras that generate the images adapted form Hartley & Zisserman's "Multiple View Geometry in Computer Vision" This tasks and algorithms as well as the methods which allow us to reason about the quality of our results are the core of the lecture 3D Computer Vision II. This lecture deals with multiple view geometry problems and having attended 3D Computer Vision I is of great benefit, however, it is not, in any case, compulsory.


Edit | Attach | Refresh | Diffs | More | Revision r1.80 - 10 Oct 2007 - 22:35 - SelimBenhimane

Lehrstuhl für Computer Aided Medical Procedures & Augmented Reality    rss.gif