Syllabus
CSSE 463 – Image Recognition

General Course Information

Course catalog description

Introduces statistical pattern recognition of visual data; low-level visual feature extraction (color, shape, edges); clustering and classification techniques. Applies knowledge to various application domains through exercises, large programming projects in Matlab, and an independent research project. Familiarity with probability distributions will be helpful, but not required. Prerequisites: Junior standing, MA221 and programming experience (e.g., CSSE220, ME323, ECE480).

Less-formal description

This course looks at image recognition, or image understanding, the process of extracting useful information out of images to make decisions about the world. Examples include photo organization and retrieval, video surveillance, and fingerprint recognition. It uses image processing and pattern classification techniques, and in some ways is the intersection of these fields. A typical week will include 1 day of "lab" work, implementing and using algorithms using Matlab. The weekly schedule can be found here.

Meeting time and place

Instructor

Matt Boutell – Assistant Professor of Computer Science and Software Engineering

Email: boutell@rose-hulman.edu
Office phone: 812.877.8534
Office address: Moench F-224
Home page: http://www.rose-hulman.edu/~boutell
Office hours: I am usually in my office every day from 8 a.m. until 5 p.m. unless I'm in class or a meeting. Please stop by whenever. If my door is shut, just knock.

Teaching Assistants

Resources

Optional text

Image Processing, Analysis, and Machine Vision by Milan Sonka, Vaclav Hlavac, Roger Boyle

Cengage-Engineering; ed. 3 (2007). ISBN-10: 049508252X, ISBN-13: 978-0495082521


Other texts I've used in the past:
  1. Computer Vision: A Modern Approach by David Forsyth and Jean Ponce

    Prentice-Hall, 2003. ISBN 0-13-085198-1

  2. Computer Vision by Linda Shapiro and George Stockman

Matlab Programming

Course Materials

I'll use Angel to post grades, dropboxes to submit labs and homeworks, and links to the course schedule, assignment descriptions, slides, and handouts. All the linked materials are available via any of the several mechanisms for accessing Public AFS data. Thus, you can get to the majority of course materials:

Learning Outcomes

Students who successfully complete this course should be able to:

  1. Describe and explain the difference between various color spaces, such as RGB, HSV, and Ohta.
  2. Compute morphological operations on simple image elements and use them to aid object recognition.
  3. Describe and implement edge detection.
  4. Use shape features for object recognition:
    1. Compute compactness ratios for various circles, rectangles and triangles.
    2. Compute or describe the computation procedure for the covariance matrix of an image element, as used to determine principal axes, and plot major and minor axes given a set of eigenvectors.
  5. Interpret an intensity histogram and compute an optimal threshold from probability density functions of the foreground and background.
  6. Describe principles of classification: feature space, decision rules, decision surface. Draw a plot of tabulated feature data to represent a 2D feature space, and specify class center positions and classification boundaries.
  7. Apply a classifier such as a neural network or support vector machine to identify images and regions.
  8. Implement a data clustering algorithm such as k-means, and apply it to image segmentation.
  9. Perform basic laboratory tasks accurately, learn to use an appropriate image processing software tool, and submit labs and homework assignments that are clear, concise, and informative, and conform to standard writing guidelines.

Homework and Projects

Grading

I plan to use the weighting scheme shown in the table below when assigning final grades.

Criteria Weight
In-class Quizzes 10%
Lab Assignments 20%
Projects 20%
Term Project 25%
Midterm Exam and take-home quizzes 25%

Letter Grades

I will do my best to conform to the Rose-Hulman definition of the various grades, as described in the Academic Rules and Procedures. Note in particular that the phrase ”thorough competence to do excellent work“ appears there in the description of the “B” grade, and it further states that “B” and “B+” will not be given for mere compliance with the minimum essential standards of the course.

Citizenship Counts!

I may adjust your overall average up or down by up to 5 points, based on your class citizenship. This includes attendance, promptness, preparation for class, positive participation in class, constructive partnership in labs and projects, and timely completion of various surveys.

The in-class time in this course constitutes an important learning experience. Unexcused absences will affect your citizenship grade. After three unexcused absences, you must discuss continuation of the course with the instructor. Four unexcused absences will result in automatic failure of the course.

Late Policy

Each student starts the term with two "slip-days" in his/her account. One slip-day, used on any lab, project, or homework assignment (unless specified otherwise), grants a 24-hour extension. It is up to the student to turn in work within that time frame, if it falls on a non-class day. Moreover, you may deposit a slip-day by turning in an assignment at least 24 hours early. Only one slip day may be withdrawn or deposited on any given assignment. Slip days may not be used on daily quizzes. Students should fill out the Angel survey (in the drop box folder) when depositing or withdrawing a slip day. Once the slip day period is past, no credit will be given for any assignment.

Email

I usually check my email several times per day (including some evenings), and do my best to respond quickly. It is a good way to get answers to simple questions. I expect you to check your email daily (not necessarily on weekends, although even that is not a bad idea). When I send mail to you, I will use your Rose-Hulman address. If you do not currently read mail that is sent to that address, please have it forwarded to wherever you read mail.

Electronic Distraction Policy

I do my best to keep class interactive. But I recognize that sometimes it is hard as a student to stay focused on the class. With laptops in class, there are many more ways to become distracted. Unfortunately these distractions are much more pernicious, since it is very easy to get drawn into things like IM conversations or RSS feeds.

I strongly encourage you to turn off IM and email software and only use other software for things directly related to class. If you choose to use non-class-related software during class, then you must sit in the back row. Doing so will prevent your classmates from being distracted by what is on your screen.

Integrity

Recall the Institute policy on academic misconduct:

“Rose-Hulman expects its students to be responsible adults and to behave at all times with honor and integrity.”

Exams, written homework, and some labs and projects will be done on an individual basis. The simple rule of thumb for individual work is:

Never give or use someone else's code or written answers.

Such exchanges are definitely cheating and not cooperation. The departmental statement on academic honesty has more detailed advice.

I encourage you to discuss the problems and general approaches to solving them with other students. However, when it comes to writing code, it should be your own work (or the work of your group if it is a group or partner assignment). If you are having trouble understanding how some library code works or pinning down a run-time or logic error in your program, by all means talk to someone about it.

If you use someone else's ideas in your solution (or any other work that you do anywhere), you have to:

If you are ever in doubt about whether some specific situation violates the policy, the best approach is to discuss it with your instructor beforehand. This is a very serious matter that I do not take lightly. Nor should you.

In general, you should not look at another student's code to get ideas of how to write your own code. Beginning the process of producing your own solution with an electronic copy of work done by other students is never appropriate.

Plagiarism or cheating will result in a negative score (i.e., less than zero) for the assignment or exam. Egregious cases will result in a grade of "F: for the course. More importantly, such dishonesty steals your own self-esteem. So don't cheat.

Developed by Matt Boutell, format and lots of wording courtesy of CSSE120 instructors.