Image Processing ECE 481
Homework 2
M.N. Wernick
Due February 1, 2016, 8pm Chicago time
1. Chroma Key (MATLAB problem)
Goal
In this problem, you will write a MATLAB program that takes an image of a person
in front of a g
ECE 481
Image Processing
Chapter1. Introduction
Dr. Joohee Kim
ECE Image Processing, Spring 2015
Course Overview
Instructor
Dr. Joohee Kim (Siegel Hall 329, [email protected])
Office hours: Tue/Thu 11:25 AM~12:40 PM
TA
Zhengyu Xia ([email protected])
O
HWG
solutions
cm
to
MATLAB
my
the
see
study
2.
a
)
the
=D.g
)
LH
program
results
code
problem
and
.
ttftttm ( 11017 c )

.
g'=e=Ffegzgtf+I+rEQto

me
)
?g+ziEeE+2QIQE
2Q=
CHIE
.toIlI=g

fb)
Same
c)
=
d)
form
YX
enforces
goals
small
,
=

=
=
class
in
so
SOLUTIONS Homework 5
1.
a)
b)
c)
d)
e)
f)
g)
h)
i) Here, erasersize = 35 (homework had a typo, saying to use 100)
j)
k) What we learned:
a) Low frequencies
% Solution  problem 1, Homework 2
%
% Chroma key compositing algorithm.
% Replaces a green screen background in an image with a new background.
%
function image_out = chromakey(fname_main, fname_bkg, thresh1, thresh2);
%
% Inputs:
%
fname_main = characte
Tomography
Tomography means imaging of the inside of
an object from measurements taken externally
Comes from Greek word tomos = slice (graphy means picturetaking)
Tomography is a kind of computed imaging
can only make the images by solving an
inverse
Image Processing ECE 481
Homework 8
M.N. Wernick
Due April 30, 2016 8pm CST
1. Huffman coding. Consider the following source (string of data) whose possible
values are 0,1,2,3:
0 0 1 2 0 0 0 3 2 0.
a) From this example data, estimate the probability for e
Image Processing ECE 481
M.N. Wernick
Homework 6
Due date: April 11, 2016 8pm CST
_
1. Computer assignment: Image restoration
The file g.tif contains a noisy, blurry image of a license plate the blur is a motion blur. The message on
the license plate tell
function H = reconfilter(N,w)
%
%
%
%
Creates a 1D reconstruction filter (combination of ramp and Gaussian) of
length N with Gaussian width parameter w.
Setting w = 0 tells the function to only use the ramp filter (no
Gaussian)
% create ramp and gaussian
%In this solution I don't bother zeropadding because the license
%plate already has a large blank border and I don't care if
%the background area is correct.
N=256;
0imension of image
Snn=10;
D=32; %Length of motion blur kernel
%Create psf h for motion b
Solutions
I
HW ?

.
y
"
s
"y
\
,
"
.at
t.my#oziitim
a)
line
The
is
it
at
integral
or
onto
ID
a
.
projects
'
x
=
cost
.
For
the
Therefore
Pocx )=J(x
'
of
'

coso
angle
,
)
S
ZD
a
'
O
axis
,
b)
Iocw
)
It
=
(w,O )
)=etwos0
Iofy=Fcfw_
=Fcfw_
Scx :c
pocxi )
%Homework 7  FBP reconstruction
N = 128;
%Problem 2  no noise case
load sinograms;
%read sinogram file
[fhat,m]=fbp481(sino,0);
%reconstruct image
%Save the nonoise result, because we will pretend it is the true image
% in part c
f = fhat;
%Problem 2c
ECE 481
Image Processing
Chapter2. Digital Image Fundamentals
Dr. Joohee Kim
ECE Image Processing, Spring 2015
In Chapter2
Several concepts related to digital images and some
of the notations will be introduced:
Image sensing and acquisition
Image sampli
ECE481 Image Processing (Spring 2015)
Homework #3
Reading assignment: Chapter 4 (Gonzalez and Woods, Digital Image Processing)
1. Show that the Fourier transform and its inverse are linear processes.
[Answer]
2. Let F (u , v ) denote the DFT of an image f
ECE481 Image Processing (Spring 2015)
Homework #1
(Due date: Feb 5, 2015)
Reading assignment: Chapter 2 (Digital image processing by Gonzalez and Woods)
1.
(20 points) A common measure of transmission for digital data is the baud rate, defined as the
numb
ECE481 Image Processing (Spring 2015)
Homework #3
Reading assignment: Chapter 4 (Gonzalez and Woods, Digital Image Processing)
1. Show that the Fourier transform and its inverse are linear processes.
2. Let F (u , v ) denote the DFT of an image f ( x, y )
ECE481 Image Processing (Spring 2015)
Homework #1
(Due date: Feb 5, 2015)
Reading assignment: Chapter 2 (Digital image processing by Gonzalez and Woods)
1.
(20 points) A common measure of transmission for digital data is the baud rate, defined as the
numb
ECE 481 Image Processing
Spring 2015
Instructor: Prof. Joohee Kim
Office: Siegel Hall 329
Phone: 3125678649
EMail: [email protected]
Office hours: Tue/Thu 10:00~11:00 AM SH 329
Time and Place: Tue/Thu 11:25 AM~12:40 PM, LS 121
TA:
Zhengyu Xia ([email protected]
function [fhat,m] = fbp481(sino,w)
%Reconstructs an NxN image fhat from an NxN sinogram p.
%N is detected automatically. Displays each step in the algorithm
0or illustration purposes. Returned variable m is a movie of the results.
%Establish dimensions of
Image Processing ECE 481
Homework 7
M.N. Wernick
Due April 30, 2016 8pm CST
1. Tomography (math)
In this problem, we will do by hand what is done in a medical imaging scanner. In part
(a), you will compute the projections of a simple object; in part (b),
Image Processing ECE 481
Homework 5
M.N. Wernick
Due date: March 23, 2016 8pm CST
_
1. Computer assignment: Some image filters
In this problem, you will apply that program to try out some image filters. You can use your own
convolution program from the pr