Purpose of Course:To introduce the topic of Computer
Vision to graduate students.
The course will discuss all three levels of computer vision: early processing,
mid--level vision or feature extraction, and high--level vision or
recognition. We will cover some basic material, but will emphasize
state-of-the-art techniques and advanced applications.
Text:New Computer Vision Textbook, by Shapiro and
Stockman. Hot off the presses, in fact, it hasn't even gotten to the
real presses. All finished chapters will be handed out, and are
available online. Comments are welcome.
Also, both Vol. 1 and Vol. 2 of the 1992 Haralick and Shapiro text
Computer and Robot Vision
should be on reserve in the library as is
the Jain, Kasturi, and Schunk text, Machine Vision. This latter
text is good reading for students who have little or no previous
background in computer vision / image processing.
Doug is putting together a set of C programs for some important
low-/mid-level operations that we often use in 3D object
recognition. These will available for downloading from the course web
and use on PCs. Students should be able to write their own C/C++
programs to add additional capabilities as needed.
Project:Each student will propose, design, and implement a
program that does some kind of machine vision. Possible topics and
more specific requirements are available here.
The project is geared to take 4-5 weeks and requires a 5-10 page
report describing the program and the results.
Topics to Cover:
We will talk about what to skip, what basics to cover, and
what advanced applications are of most interest. The chapters
of the text and their present states are given below as a guide:
* = not yet written
- Imaging and Image Representation
- Binary Image Analysis
- Pattern Recognition Concepts
- Filtering and Enhancing Images
- Content-Based Image Retrieval
- Image Segmentation*
- Matching in 2D
- Perceiving 3D from 2D Images
- 3D Sensing+
- 3D Models and Matching
- Virtual Reality
- Related Topics*
+ = not yet finished, but coming along well
The grades will be based on a combination
of small homework sets, midterm, final, and project.
Projects will be due on the last day of class, so I have ample
time to grade them.