Sina Farsiu
BME 271: Signals and Systems (Spring 2015, Spring 2016, Spring 2018)
BME
544 - 01:
Digital
Image
Processing
(Spring 2011,Spring 2012, Spring 2013, Spring 2014, Fall 2017, Spring 2019, Fall 2020)
TENTATIVE COURSE
INFORMATION:
Instructor: Prof. Sina Farsiu
Time: Wed-Fri 10:05-11:20 AM,
Permission code: Please directly contact Ms. Susan Story-Hill (sshstory@duke.edu) to receive permission code.
Room: TBD
Course Description: Introduction to the theory and
methods for digital image sampling, enhancement,
visualization, storage, reconstruction, and analysis with
emphasis on medical applications. This course is mainly
designed for BME graduate students. However, graduate
students from all engineering disciplines, computer science,
and senior undergraduate students who have already passed
required courses may also participate.
Prerequisites: The student must have passed one
undergraduate course on signals and systems and one course
on probability and statistics:
A (Probability and Statistics):
MATH 135 or STA 113 or ECE 255 and EE 64 or permission of
instructor.
B (Signals and Systems): BME 571
or ECE 54 or permission of instructor (if you find this
comic strip funny " link",
you probably know enough about signals and systems).
The students must also have a
basic knowledge of the MATLAB software.
Text: Digital Image Processing, XX
edition, by R. Gonzalez and R. Woods, 20XX. ISBN
number XXX. Although,
several lectures are loosely based on the text book
material, for which handouts and journal articles will be
provided by the lecturer.
Student
Evaluation (MORE DETAILS IN THE BLACKBOARD POSTINGS): Homework (15%), a midterm exam
(35%), and a final project (50%).
Midterm
exam will include a one-to-one interview with the professor.
The material covered after the midterm exam might be
questioned at the final presentation. The final project will
include a class presentation at the end of the semester, an
unlimited page report due one week before
the-end-of-the-term presentation, and one four-page paper
following the style of the IEEE International Conference on
Image Processing (ICIP) papers due on the day (and in lieu)
of the final exam. Final projects can be done individually
or in a group, however, each student must have a defined
role approved by the professor. Professor will help in
project selection. Homework, mfiles, and reports must be
submitted electronically by 10:00AM of the deadline unless stated otherwise by the lecturer(late submissions are not
accepted).
Course Objectives: The student will gain a basic knowledge of the most fundamental issues as well as novel topics in image processing. By the end of this course, he/she should have a comprehensive knowledge of an image processing topic, based on the final term project of his/her choice, and should be able to take over the image processing tasks in his/her academic career with minor required supervision.
Tentative
Course Outline:
Lecture
1
Introduction, history, applications, and fundamentals of
Image Processing.
Lectures
2-4
Spatial Domain Image Enhancement: Denoising and Contrast
enhancement
Lectures
5-7
Fourier Domain Image Representation and
Enhancement
Lectures
8-10
Registration:
Optical Flow and Phase-Based Motion Estimation
Lectures
11-16
Inverse Problems (Wiener filter,
Least-squares, Denoising, Deblurring, Blind Deblurring, Back
Projection, and Reconstruction)
Lectures
17-18
Interpolation
(Single-Frame, Multi-Frame, Super-Resolution)
Lectures 19-22 Multi-Resolution Representation, Wavelets, Sparse Representation and Compressive Sensing
Lectures 23
Biological Image
Processing: Vision and Art
Lectures
XXX Student presentations of final projects
(in case of high enrollment in this course, some
presentations will be scheduled for the last two weekends).
Last lecture:
The real world is not fair and in general
cannot be modeled as "linear", or "Gaussian", and how to
deal with it.