P33ZVD - Introduction to Computer Vision

Winter 2006/2007, 2 hours lecture + 2 hours exercises per week, i.e., 90 minutes lecture weekly + exercises.

The lecture is in the lecture room G205, building G at the Karlovo namesti campus of CVUT on Fridays from 9:15 till 10:45. Click here for the map.

Lecturer  Professor Vaclav Hlavac
Center for Machine Perception, Department of Cybernetics
Faculty of Electrical Engineering, CTU Prague
hlavac@fel.cvut.cz, http://cmp.felk.cvut.cz/~hlavac

Instructor of the exercises    Prof. Vaclav Hlavac
Plan of the lectures

No Date Lecture content Electronic material
1 05.10.2006 Participation at the one-day-colloquium. This is a whole day scientific workshop related to computer vision.  
2 12.10.2006 What is computer vision? Low-level image processing and high-level vision. Digital image. Image formation, acquisition. Digital image (cz). Image formation (cz), geom. optics (cz), cameras (cz).
3 20.10.2006 Image analysis as analysis of a 2D signal. Fourier transform rehearsal. Linear integral transformations, image preprocessing in frequency domain.. Fourier tx 1D. Fourier tx 2D (cz). Fourierova filtrace. FFT.
4 27.10.2006 Image preprocessing in image domain. Marr's theory of edge detection. Scale space. Preprocessing in the image domain. Edge detection.
5 03.11.2006 Mathematical morphology. Math. morphology. Binary (cz), gray scale (cz).
6 10.11.2006 Basics of pattern recognition. Image segmentation. Basics of PR (cz). Segmentation (taxonomy, thresholding, spatial coherence).
7 17.11.2006 No lecture. Public holiday.  
8 24.11.2006 Description of objects in images. Detection of distinguished primitives in images. 2D image description.
9 01.12.2006 Texture. Color.  Color.
10 08.12.2006 3D vision geometry. A single camera and more cameras. A single camera. More cameras.
11 15.12.2006 Correspondence problem. Reconstruction of 3D scenes.  
12 22.12.2006 Motion analysis. Motion analysis
13 5.01.2006 Image and video compression. Image compression. Video compression.
14 12.01.2006 Consistent labeling (research talk, lecturer T. Werner)  

Exercises

The exercises have two goals.

  1. I like to teach a PhD candidate how to write a scientific paper and prove it on a domain related both to student's own research and computer vision. The student either works on his own research paper to be submitted elsewhere the topic of which is related at least loosely to computer vision. If the student's own research is too far from this subject then he will be given a topic which is expected to be studied and described in a research report. The text has to be written in English. After the topic is agreed with the instructor, the students works on it individually. However, the student is recommended to conduct personal consultations with the instructor regularly.
  2. The student has to acquire some practical skills with computer vision algorithms. The student has to develop one module in Matlab which will contribute to the cmpvia. This is a public domain toolbox intended to help student in their self-study. The documentation and comments in the code have to be written in English. Click here for more details.

Requirements for the credit: approved paper and one developed and approved module to cmpImageAnalysisToolbox.

Examination: Written test + oral discussion.

Note: We agreed on 11.10.2006 to move the lecture to Fridays, 9:15 because it is convenient to all student.

Maintained by V. Hlavac, hlavac@fel.cvut.cz
Last modification 04.04.2007 16:10