Image Analysis with Statistical Models



Instructors: Joachim Buhmann; Wolfgang Einhäuser
TA: Björn Ommer

Lecture (Vorlesung): Wednesday, 9-11 (CAB H52)
Recitation (Übung): Wednesday, 11-12 (CAB E38)

Overview
In recent years, statistical methods and models have proven very successful in image analysis and understanding. The lecture covers the reconstruction of 3D objects from image data, the application of Markov random fields to image processing, and the use of graphical models for image understanding.
Administrative Details can be found here

Literature
part 1: Horn - Robot Vision. (MIT Press)
part 2: Winkler - Image Analysis, Random Fields and Markov Chain Monte Carlo Methods. (Springer)

Syllabus and Handouts (contain copyrighted material, for use in this lecture only):

Date Topics Slides/Handouts Additional Materials Homework
Sep, 26 Introduction; The Physics of Image Formation; Lecture 1
Oct, 3 Lambertian Surfaces, Surface Properties, Reflectance Maps Lecture 2 Exercise 1
Solution
Oct, 10 Photometric Stereo, Albedo, Shape from Shading Lecture 3 Hertzmann & Seitz (2003) Exercise 2
Solution
Oct, 17 Shape from Shading II, Scene Probability Equation, Generic Viewpoint Assumption Lecture 4 Freeman (1994) Exercise 3
(Code)
Solution
Oct, 24 Scene from Texture Lecture 5 Clerc & Mallat (2002) Exercise 4
(images)
Solution
Oct, 31 Markov Random Fields I: Cleaning Noisy Images Lecture 6 Geman & Geman (1984) Exercise 5
(Images)
Solution
Nov, 7 Markov Random Fields II: Formal definition; equivalent formulations of MP Lecture 7 Exercise 6
Solution
Nov, 14 Equivalence of Markov Properties Lecture 8 "
Nov, 21 Sampling, Markov Chains, Contraction Coefficient Lecture 9 Exercise 7
(Solution)
Nov, 28 Gibbs Sampling, Annealing Lecture 10 Exercise 8
Solution
Dec, 5 Graphical Models I: Bayesian Networks Lecture 11 Exercise 9
(Solution)
Dec, 12 Object Categorization using Graphical Models Lecture by B. Ommer
Dec, 19 Inference on Graphical Models Lecture 13
Most of the slides were created and kindly provided by V. Roth. Slides contain copyrighted material from various sources and are intended for use in the course only.
Last modified: Tue May 15 09:47:06 MEST 2007