TeachingSs06PSModernApproach

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

Proseminar: Basic Principles of Computer Vision

Nassir Navab, Andreas Hofhauser, and Darko Zikic

SWS: 2, ECTS: 4 Credits
Time: 08 Feb 2006 (Thursday), 10:30-14:00 in the seminar room 03.13.010.
Location: Seminar room 03.13.010

The primary course language is English

Schedule

  • 16 Feb 2006 at 10:00. The compulsory preliminary discussion in room 03.13.010.
  • Until 01 Apr 2006 : First discussion of the topic with the respective tutor (please contact the tutor for individual date):
    • Given literature should be worked through before this meeting and the basics of the subject well understood.
    • Prepare also eventual techical and detailed questions regarding the subject so we can discuss them.
    • Plan for the talk and paper should be ready before the meeting so it can be discussed.
  • Until 24 Apr 2006 : Second discussion with tutor (please contact the tutor for individual date):
    • Final versions of talk and presentation are supposed to be ready by now. Send them to the tutor prior to meeting so that they can be discussed at the meeting.
  • 01 May 2006 : The very final version of paper and presentation (including eventual changes from the second discussion) is to be mailed to the tutor.
  • Presentation Day : Somewhere between 01 and 08 May. We will set the actual date and time in accordance to the students' schedule...
  • Presentation: 08 Feb 2006 (Thursday), 10:30-14:00 in the seminar room 03.13.010.

Introduction

The original idea of Computer Vision is to make a computer "see", as we do. During the last few decades, computers really started to see and understand images, since the theoretical background was formulated and realized in sophisticated Computer Vision applications. Computer Vision techniques have found their way into entertainment, television, industry, medicine, broadcasting, and many other fields.
In this Proseminar we will introduce the basics of Computer Vision. We will start with the basic descriptions of the geometrical entities and transformations as these are the foundations for everything later on. Then we will discuss the properties of an image and the general principles of the formation of an image. This brings us to the properties and geometry of an off-the-shelf camera. We will look at how such cameras can be calibrated and their images be undistorted, or how two cameras can be combined to a stereo system, which can be used to recover the depth of an arbitrary scene, or to immerse into a scene wearing a "virtual reality helmet" as in computer games. Finally we will go a step further and consider simple image processing techniques such as edge detection or image smoothing. We will also try to give an outlook on further and more sophisticated applications from CV applications from the industry, the connection of CV to Medical Imaging, as well as to Augmented Reality.

Topics

Work in progress. Hence: subject to change.

Topic Detailed Description Literature Student Tutor Handouts/Presentations
Homogeneous Coordinates and 2D Projective Space The homogeneous coordinates are crucial for Computer Vision applications and many others. In computer graphics homogeneous coordinates are used for transformations and for camera modelling.
  • Modelling of basic geometrical entities in the 2D projective space
  • Basic geometrical transformations in the 2D projective space (rotations, translations, affine transf., ...)
  • Relations between the geometrical entities in the 2D projective space
  • Hartley/Zisserman: 2.1-2.4
  • Short Intro to P2
  • Darko Zikic Darko Zikic  
    Camera Models Description of how a camera works. Useful in Computer Vision as well as in many other contexts.
  • mathematical modelling of how a camera works, i.e. how a point from 3D is transformed into the image
  • different camera classes
  • basics of camera calibration
  • Forsyth/Ponce Chapter 1/2/3
  • Hartley/Zisserman: Chapter 6
  • Alexander Sporn Darko Zikic  
    Image Filtering A very important step in order to get closer to making a computer 'see' is to be able to manipulate images and to extract certain properties from those. This can be done by filtering.
  • filtering by convolution
  • convolution: properties and computation
  • edge enhancement filters
  • smoothing filters
  • Forsyth/Ponce Chapter 7/8
  • Handbook of Computer Vision and Applications Volume 2 Chapter 5/10
  • Pierre Schroeder Andreas Hofhauser  
    Template Matching and
    Similarity Measures
    Here we take a look at basic methods of finding a representative model of an object (patches/templates/patterns) in larger images. For this we need a method of comparing a similarity of two images, the so called similarity measure.
  • similarity measures: SAD, SSD, CC, NCC
  • template matching
  • Dr. Carsten Steger. Lecture "Methoden der industriellen Bildverarbeitung"
  • William L. Briggs and Van Emden Henson. "The DFT: An Owners' Manual for the Discrete Fourier Transform", pp.78-86
  • Wolfgang Wein. Diploma Thesis: Intensity Based Rigid 2D-3D Registration Algorithms for Radiation Therapy
  • Pavlo Kerestey Andreas Hofhauser  

    Registration / Preliminary Discussion

    To register for the Proseminar, please send an email to Darko ().

    The compulsory preliminary discussion will take place on thursday, 16 Feb 2006 at 10:00 in room 03.13.010.

    Requirements

    • the presentation should be about 45min. One session, i.e. presentation + discussion during the presentation should take about an hour.
    • the handout of 8-10 pages (in LaTex)
    • the outline should be send to the supervisor one month before the presentation
    • the presentation and the handout are be finalized and discussed with the supervisor at least one week before the presentation so that eventually changes can be made.

    Literature

    Templates

    Useful Links


    TeachingForm
    Title: Basic Principles of Computer Vision
    Professor: Nassir Navab
    Tutors: Andreas Hofhauser, Darko Zikic
    Type: Proseminar
    Information: Proseminar, 2 SWS, 4 ECTS
    Term: 2006SoSe
    Abstract:  


    Edit | Attach | Refresh | Diffs | More | Revision r1.33 - 02 Jun 2006 - 10:28 - DarkoZikic

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