




Winter 2011 
ECE 278a 

Instructor: B. S. Manjunath 



Overview
This course provides
an overview of the fundamental topics in image processing. Lectures emphasize
the basic theory and the course projects explore some current topics and
emphasize algorithm development and implementation. Course outline and
additional information distributed in class on 01/04/2011 (Course Outline) Topics
for Winter 2011
1. Introduction: an overview of some
of the preliminaries. a. Digital picture transforms, 2D
DFT. b. Random fields and statistical
image models. c. Image representations: 2D
sampling theory, generalization to random fields, representation using
orthonormal basis functions. 2. Digital Image Transforms and
Coding a. Overview of transform compression b. Various transforms: KarhunenLoeve, Fourier, Cosine, Hadamard,
Walsh, É c. 2D Wavelet transforms d. JPEG and JPEG2000. 3. Image Restoration a. Inverse and Leastsquares
filtering b. Maximumlikelihood (ML) and
Maximum aposteriori (MAP) estimation 4. Image Reconstruction from
Projections a. Image Projections and the Fourier
Slice Theorem b. Filtered back projection
algorithm c. Iterative reconstruction
techniques 5. Selected Topics: these will cover
the general areas of image segmentation and image registration. Papers and
course projects will explore these topics in more detail. Grading
30% for H/W, preparing critiques,
paper presentations, and participation in discussions; 30% for the class
project; 40% for the final. HWs are due in class on the due dates. Class
projects should involve some kind of computer implementation. Individual projects
strongly encouraged but groups of two ok if the project needs are justified Text/References
There is no required text book for the class. Lecture notes will be posted
(will try my best to post them before the lectures). However, not everything
that is discussed in class will be in these notes. I will also handout
additional reading materials during the quarter. 

Important dates Jan 13, 2011 Paper assignments, groups
January 25 Project proposals are due on or before this date. Include a brief summary of proposed work and any appropriate references. Proposals should be no longer than 2 pages and hardcopies only. I will return the proposals with my comments on February 1. You are strongly encouraged to meet with me and discuss your proposals before submitting it.
Feb 3: Paper Critiques Due Before Class (email the electronic version, PDF files only, that will be posted on the class web site).
Presentation schedule (plan for ~35 mins presentations and 10 mins Q&A) Distribute the critiques by Feb 3 (all topics).Feb 8: (a) Compressed sensing [slides] (b)Face recognition [slides] Feb 10: (a) Image Forensics [slides], (b) Mobile Image Processing [slides] March 810 Inclass presentations (approx.. 15 minutes) of your project. March 11 Final report due by 5PM March 15, 123PM Final Examination Homeworks
HW#4 (due Feb 15, by email) Use the following images in your experiments, note that one of them is color. You can either process the gray scale image or work in color (extra credit). Temple, Monkey. 2007 Exam with solutions (this was a takehome exam) 2009 Exam (solution) (this was also a takehome exam)

Lecture Notes
Lecture 1 Lecture 2 Lecture 3 Lecture 4
Lecture 05 (handout on digital transforms part1 part2) Lecture 07 Paper by Burt and Adelson on Laplacian pyramids (required reading). Additional Reading: Multiresolution representation using wavelets by Mallat. (pdf) Lecture 09 ( see also Figures from G&W Chap 7 and a brief overview of JPEG compression) Lecture 10 [new: MSE derivation from Prof John Shynk] Lecture 11 [student paper presentations][compressive sensing] [face recognition using sparse representations] Lecture 12 [student paper presentations][image forensics][mobile image processing] Lecture 13 [deblurring examples in Matlab][examples of inpainting][review of histogram equalization] Lecture 16 (meanshift clustering) Lecture 17 (Level set segmentation Notes) Lecture 18 (continue Level set discussions and wrapup)
Note on Image Registration: Many of you are working on the image registration project. You may find the lecture notes developed by Marco Zuliani very useful.




