Computer Vision Course

 

CMSC 828D: Fundamentals of Computer Vision (Fall 2000)

Mondays and Wednesdays, 2:00 – 3:15 p.m.  A. V. Williams 1112

 

Larry Davis, Ramani Duraiswami, Daniel DeMenthon, and Cornelia Fermüller

 

Send mail to the entire class.

 

This course will provide an advanced introduction to computer vision while emphasizing basic techniques from numerical analysis and applied mathematics (linear algebra, probability and statistics, optimization). Our goal is to provide every student who takes the courses a basic set of tools to read technical papers, and pursue research in the field.

 

This course will count for one of the ten courses for the Ph.D. requirement. (See the graduate catalog for more details)

 

Check administrative notices at the bottom of this page.

 

Grading: Combination of attendance, homework and exams.

 

Attendance and homework submission are mandatory.

 

Homework: There will be weekly homework emphasizing completion of derivations, problems, and short programs in Matlab. Drop worst homework. (50 %). Homework will be handed out at the end of class Wednesday and due Monday before class.

 

Exams: Two examinations. midterm and final each 25 %

 

Books: No prescribed text.

However, the following books will be useful

  1. Introductory Techniques for 3-D Computer Vision, by Emanuele Trucco, Alessandro Verri, Prentice-Hall, 1998.
  2. A Guided Tour of Computer Vision, by V. S. Nalwa, Addison-Wesley, 1993.
  3. Multiple View Geometry, by Richard Hartley, Andrew Zisserman, Cambridge University Press, 2000.
  4. Numerical Recipes in C, by William Press et al., Cambridge Univ Press, 1992.
  5. Pattern Classification and Scene Analysis, by Richard O. Duda, Peter E. Hart, John Wiley & Sons, 1973.

 

Matlab is the programming environment of choice. A short introduction to Matlab for computer vision can be found here. (this was handout 1)

 

A web site with lecture slides, homework and solutions will be posted online.

 

Tentative Sequence of Lectures (subject to change)

1

08/30/00

Introduction. Course overview. Introduction to Matlab.

Duraiswami

Homework1

Solution

2

09/06/00

Image Creation, Radiometry, Photometry, Color.

DeMenthon

Homework2
Solution

3

09/11/00

Pinhole camera, camera models, lenses. Calibration: intrinsic and extrinsic parameters. (6 slides per page version)

DeMenthon

 

4

09/13/00

Linear Algebra: Matrices, Rank, Products, Eigenvalues and Eigenvectors. (6 slides per page)

Duraiswami

Homework3

Solution

5

09/18/00

Alternative theorems Least squares, Singular Value Decomposition. (6 slides per page)

Duraiswami

 

6

09/20/00

Projective Geometry. (6 slides per page)

DeMenthon

Homework4

Hints

Solution

7

09/25/00

Basic calculus and approximation. Interpolation, Splines (6 slides per page)

Duraiswami

 

8

09/27/00

Local Features and Grouping.(6 slides per page)

Davis

Homework5

Solution

9

10/02/00

Calibration (6 slides per page)

DeMenthon

 

10

10/04/00

Pose and Rigid body motion. (6 slides per page)

DeMenthon

Homework6

Solution

11

10/09/00

Probability and Statistics. (6 slides per page)

Davis

 

12

10/11/00

Pattern recognition and learning. (6 slides per page)

Davis

Homework7

Solution

13

10/16/00

Object Recognition. (6 slides per page)

DeMenthon

 

14

10/18/00

Stereo I.(6 slides per page)

Davis

Homework8

Solution

15

10/23/00

Optimization - derivative free.(6 slides per page)

Duraiswami

 

16

10/25/00

Optimization - Newton and Levenberg Marquardt.(6 slides per page)

Duraiswami

Homework9

solution

17

10/30/00

Stereo II.(6 slides per page)

Davis

 

18

11/01/00

Mid Term Exam (closed book).

--

Solution

19

11/06/00

Multi-camera and long baseline stereo (volume intersection, space carving)

Davis

 

20

11/08/00

Shape from Shading.(6 slides per page)

DeMenthon

Homework10

solution

21

11/13/00

Motion and Flow I(6 slides per page)

Fermuller

 

22

11/15/00

Motion and Flow II(6 slides per page)

Fermuller

Homework11

solution, Matlab

23

11/20/00

PCA, MLE, and MAP.(6 slides per page)

Duraiswami

 

24

11/22/00

Tracking: Motion models, Kalman Filter.(6 slides per page)

DeMenthon

Homework12

solution

25

11/27/00

Tracking humans from video..(6 slides per page)

Davis

 

26

11/29/00

Sampling Methods. Condensation Tracker..(6 slides per page)

Duraiswami

 

27

12/04/00

Epipolar Geometry and the Fundamental Matrix..(6 slides per page)

DeMenthon

 

28

12/06/00

3D Reconstruction from Multiple Views..(6 slides per page)

DeMenthon

 

29

12/11/00

Introduction to MPEG..(6 slides per page)

Duraiswami

Final Exam

solution

 

12/13/00

Take home final exam due to be returned.

--

 

 

Notices:

09/28/00: The solution for homework 3 has been posted.

09/27/00: The links for lecture 9 and lecture 10, and for homework 6 will be activated after those classes are over.

09/21/00: Solutions for Homework 1 and 2 have been posted. Hints for Problem 6 of homework will be posted and mailed shortly.

09/13/00: The grader for this course is Mr. Ahmed Elgammal (elgammal@cs.umd.edu, Room 3364 AV Williams). Please check with him if you have any questions on graded homework.

09/13/00: The course notes are now supposed to be only accessible from university computers. If this is a problem for you, let me know, so that alternative arrangements can be made.

09/06/00: For ECE students: To get credit for the course you need to get your advisors signature. Please see the following message from Prof. Tits.