CS5814 - Digital Picture Processing
Fall 2007, CRN 96310
TuTh 3:30pm, Shultz 105

Class Materials URL: http://pixel.cs.vt.edu/courses/cs5814.html

Instructors

R.W. Ehrich, KW2 129, Office Hours: M,W 1:00-3:00 in KW2-129, or by appointment
Seung In Park, Office Hours: W 10:00 - 12:00 in McB 133A, or by appointment

Objectives

Digital picture processing involves the acquisition, enhancement, display, and understanding of digital images. The field dates back to the work of Roberts at MIT in the late '50s and has evolved slowly into the AI image understanding systems being investigated today. This course covers what is generally called low level vision and provides a survey of the most well-known techniques in the field. The approach in the course will focus on spatial rather than on frequency domain techniques. However, some frequency-domain concepts will be introduced where that is essential to understanding ideas such as sampling and filtering.

A scan of the course text will confirm that the course is not intended to focus on signal processing. On the other hand, without mathematics it would be difficult to understand many of the ideas and techniques that are basic to the field. Although concepts such as complex numbers, convolution, norms, inner product spaces, and orthogonal expansions will be reviewed, these ideas should not be completely new to those taking the course. The Petrou text in the course references is a particularly good introduction to many of the mathematical preliminaries, and some of the lecture material will be taken from this text.

The references on reserve in the library are especially important to the course since they represent important recent topics that are not covered in course texts. Students will be responsible for reading the relevant papers.

Grading

The semester grade will be based upon assignments issued in class (15%), one midterm (25%), three programming projects (30%), and a comprehensive final examination (30%).

Homework

Homework assignments will be posted on the class web page and will be due by 5pm on the assigned dates. NO LATE HOMEWORK WILL BE ACCEPTED unless by prior approval. All homework solutions must be legible, well-reasoned, and accurately expressed. All work submitted for a grade must be the student's own work. If you have a question about the grading of an assignment or an exam, you should speak with the cource instructor within a week after the assignment is returned.

Exams

Midterm, Tuesday, September 25
Final, Friday, December 7, 10:05am - (NOTE: Classes end Wednesday, December 5)

No missed exams will be accepted unless by prior approval. All exams are closed-book exams. The final covers all the material in the course.

Ethics

The Honor Code will be strictly enforced. It is a violation to represent joint work as your own or to let others use your work; always acknowledge any assistance you received in preparing work that bears your name. You are expected to work independently unless explicitly permitted to collaborate on a particular assignment. It is not a violation to discuss approaches to problems with others; however, it IS a violation to use wording or expressions in your assignments that have been written by others without acknowledging the source.

Please be considerate of your fellow students. In particular, please realize that your classmates will likely need the reserve materials at the same time you do... don't keep these materials out for long periods of time, especially before exams and assignment due dates.

Prerequisites

Undergraduate calculus, linear algebra, probability theory, and programming competence in C or C++. For example, students should be familiar with the idea of orthogonal bases for discrete or continuous spaces.

Projects

The problem of low level image understanding is hard, and the course would not be complete without some applied work to complement the lectures. Images will be provided, and students may work on the platform of their choice, realizing that their work must be demonstrated when complete.

Text

Gonzalez, R.C. and Woods, R.E., Digital Image Processing, Prentice Hall, 2008.

References

Course Outline


        1       Visual Perception, Sampling and Quantization, Digital
                Topography, Regions, Boundaries, Connected Components
        2       Sensor Architectures
        3       Spatial Image Enhancement
        4       Linear Systems, Convolution, Correlation, Impulse Response
        5       Inner Products, Orthogonal Transformations, Introduction to
                the Fourier Transform
        6       Frequency Space Enhancement
        7       Matched Filtering, Sampling
        8       Color, Color Quantization
        9,10    Image Morphology
        11      Segmentation, Relaxation, Hough Transforms
        12      Surface Approximation
        13      Representation and Description
        14      Image Texture Analysis and Description