Probabilistic Methods in Image Analysis
Lecturer:
Dr. Bernhard Burgeth
Office Hours: Tuesday: 15:00-16:00.
Winter term 2008/09
Lectures (3h) with exercises and assignments (1h);
6 ECTS points
Lectures / tutorials:
Tuesday, 12-14 c.t., Friday, 8-10 c.t. Building E13, Lecture Hall 001
First lecture: Tuesday, October 21st, 2008
As indicated in personal e-mail messages sent out on March 26th the
2nd oral exam will take place on Tuesday, April 7th, 2009.
Please respond to the e-mail by
Only the registered participants will receive an e-mail with a detailed schedule!
Prerequisites –
Synopsis –
Tutorials –
Exams –
Course Material –
Literature
This course is suitable for students of mathematics , physics or computer
science who completed their undergraduate studies in mathematics.
Knowledge of probablity theory or statistics is
helpful but not required. The lectures will be given in English.
Hence passive knowledge of English is necessary.
Probabilistic techniques are employed quite successfully
in the processing and analysis of images, however, they also
play a vital role in pattern classification,
data mining and learning theory.
In this course we will discuss
- basic notions from probability theory and statistics as well as from
image processing
- histogram based image analysis and enhancement methods
- the probabilistic background of the Karhunen-Loeve expansion
used for data compression, for example
- independent component analysis and applications
- the notion of entropy in image registration
- and, if time permits, we will give an introduction to the basic ideas of
Markov random fields and simulated annealing.
On Fridays, 9:15 -- 10:00
The tutorials include assignments that provide important
additional mathematical insights.
The tutorials are conducted by the lecturer.
There will be an opportunity for an oral exam at the end of the lecture period, Friday, February 13th,
and a second oral exam at the end of the semester.
Active participation in the in the exercises is expected in order to be admitted
to the exams.
The better grade of the two exams counts.
Further details will follow.
Participants of the course can download the course material
(access is password-protected).
Lecture slides
Date | Topic | Lecture
|
21.10. |
Organisational Issues |
Notes |
21.10. |
Images, and Image Degradations |
Lecture 01 |
24.10. |
Elements of Probability Theory I |
Lecture 02 |
28.10. |
Elements of Probability Theory II |
Lecture 03 |
04.11. |
Elements of Probability Theory III |
Lecture 04 |
14.11. |
Elements of Probability Theory IV |
Lecture 05 |
18.11. |
Elements of Probability Theory V |
Lecture 06 |
21.11. |
Elements of Probability Theory VI |
Lecture 07 |
5.12. |
Histogram Based Operations on Images I |
Lecture 08 |
9.12. |
Histogram Based Operations on Images II |
Lecture 09 |
16.12. |
Histogram Based Operations on Images III |
Lecture 10 |
13.01. |
Parzen Windows and ML |
Lecture 11 |
20.01. |
Registration by Maximization of Mutual Information |
Lecture 12 |
27.01. |
Minimally Stochastic Approach to Diffusion Equations |
Lecture 13 |
03.02. |
PCA, POD, KLT via SVD |
Lecture 14 |
Assignments
Relevant references will be provided in the lecture.
|