Instructor: Dr. Alper Yilmaz
(yilmaz at cs.ucf.edu)
Room: CSB 250
Office Hours: 2pm3pm Monday
Wednesday
Grader: Wade Spires
(wspires at cs.ucf.edu) 
 Introduction and course objectives
 Lecture
2 (August 24, 2005)
 Imaging geometry, perspective projection etc.
 Reading from Course text book pp. 2640
 Homework assignment (due date: September 7, 2005)
 Lecture
3 (August 31, 2005)
 Imaging geometry, affine camera models, radial
distorsion
 Estimating camera parameters
 Rotation around arbitrary axis
 Homework assignment (due date: September 14, 2005)
 CORRECTION:
2nd slide rotation matrices are corrected!
 In last class and today's slides the rotations
were defines as follows:
 Rotation around zaxis is in the counterclock wise direction
 Rotation around xaxis is in the clock wise direction
 Rotation around yaxis is in the clock wise direction
 Now it is updated to be counter clock wise in all 3 axis. In
your assignments you can do it either way.
 Another correction is in the matrix notation of
the intrinsic camera parameters. ox and oy should be on the last column
of the matrix. (new)
 Lecture
4 (September 7, 2005)
 Binary, gray level and color images.
 Image noise.
 Derivatives, filtering.
 Lecture
5 (September 12, 2005)
 Blurring, noise removal.
 Image derivatives and edge detection.
 Programming assignment (due date: September 19, 2005)
 Image Viewer for windows
 Sample image
 Implementation of sobel and prewitt edge detector.
 Deliverables:
 Report: For both sobel and prewitt the report should include
blurred images in x and y directions, derivative images in x and y directions,
the gradient magnitude as image, and the edge maps for various thresholds.
 Source code.
 Lecture 6 (September 14, 2005)
 Marr Hildrett edge detector.
 Canny edge detector.
 Homework assignment (due date: September 28, 2005).
 1st Programming Project (due date: October 17, 2005)
 Implementation of Canny Edge Detector.
 Lecture 7 (September 19, 2005)
 Region segmentation
 Connected component analysis
 Reading: Chapter 3 from Dr. Shah's lecture notes
 Programming assignment (due date: September 28, 2005)
 Sample image
 Implementation of histogram based segmentation.
 Deliverables:
 Report: Show the histograms after smoothing 5 times with a gaussian filter (.05 .25 .40 .25 .05), and show the resulting binary region images.
 Hardcopy of the source code.
 Lecture 8 (September 21, 2005)
 Image segmentation: seed segmentation, region growing, region split and merge.
 Likelihood ration test and phagocyte algorithm.
 CORRECTION:
Likelihood ratio test is corrected!
 Lecture 9 (September 26, 2005)
 Region properties (segments in an image or objects).
 Area, centroid, perimeter, moments, compactness, orientation.
 Homework assignment (due date: October 10, 2005)
 Look at lecture 10 slides
 Lecture 10 (September 28, 2005)
 Derivation of orientation (elongation) of a region
 Gaussian Pyramid
 Laplacian Pyramid
 Programming assignment (due date: October 12, 2005)
 Lenna image
 Implementation of Gaussian Pyramid.
 Deliverables:
 Report: Write a report showing all the resolutions of Gaussian pyramid at least on two different examples.
 Hardcopy of the source code.
 Lecture 11 (October 3, 2005)
 MidTerm Exam 1 (October 5, 2005)
 Lecture 12 (October 10, 2005)
 Hough Transform
 Reading: Section 5.2 from text book
 Programming assignment (due date: October 26, 2005)
 Sample image
 Implementation of Hough transform. Your program
should use the one of the edge detectors you have written to find
the edges first, than apply Hough transform on the edgemap
 Deliverables:
 Report: Write a report which shows the detected edges, and the (theta,rho) voting matrix.
 Hardcopy of the source code.
 Lecture 13 (October 12, 2005)
 Hough Transform (continued).
 Circle fitting
 Generalized hough transform.
 Medial axis transform
 Interest point detectors
 Lecture 14 (October 17, 2005)
 Optical flow
 Horn and Schunk.
 Schunk
 Lucas and Kanade
 Lecture 15 (October 19, 2005)
 Global Motion
 Programming assignment (due date: November 16, 2005)
 Sample images flower garden sequence <mpeg>
 Implementation
of Lucas Kanade. Create 2 levels of pyramid and for each level compute optical flow independently.
 Deliverables:
 Report: Write
a report which shows the input images, and computed flow fields
for each pyramid level independently. You can use matlab "quiver" function to show the flow field.
 Hardcopy of the source code.
 Lecture 16 (October 26, 2005)
 Block based optical flow
 Token based optical flow
 Lecture 17 (November 1, 2005)
 Structure from motion
 Orthographic displacement model
 Perspective displacement model
 Homework assignment (due date: November 9, 2005)
 2nd Programming Project (due date: December 5, 2005)
 Implementation of Anandan's approach for global motion compensation.
 Sample test images head sequence, claire sequence.
 See slides for deliverables.
 Lecture 18 (November 3, 2005)
 Stereopsis
 Simple stereo system
 Token based stereo
 Correlation based stereo
 Lecture 19 (November 7, 2005)
 Revisit Programming project #2
 Lecture 20 (November 9, 2005)
 Revisit Generalized Hough Transform
 Revisit Motion
 Lecture 21 (November 14, 2005)
 Epipolar Geometry
 Essential Matirx
 Fundamental Matrix and their derivation
 Lecture 22 (October 16, 2005)
 Midterm Exam 2 (21 November 2005)
 Lecture 23 (November 28, 2005)
 Image segmentaion using graph cuts
 Last day for bonus assignment is Friday 4pm.
 FINAL December 7, 2005, Time 1.00pm3.50pm., Place: Regular classroom

Course Goal
This course is introductory level. It will cover the basic topics of
computer
vision, and introduce some fundamental approaches for computer vision
research.
Grading Policy (Updated 11 September 2005)
Biweeky Assignments: 20%
Programming Assignments: 20%
Programming Projects: 20%
MidTerm Exam: 20%
Final Exam: 20%
Grading: (Updated 3 October 2005)
The final grades will be given in the form of a+, a. The conversion table is given below.
100 
95 
a+ 
94 
90 
a 
89 
80 
b+ 
79 
70 
b 
69 
.. 
c 
Prerequisites: A good background in
calculus, geometry, linear algebra, programming
in MATLAB or C.
The University Golden Rules will be observed in this
class. Copying or Plagiarism is violation of the Golden Rules.


