0% found this document useful (0 votes)
32 views

Medical Imaging Techniques: Textbook and Materials

This document discusses medical imaging techniques and digital image processing. It provides an overview of textbooks and materials on the topics. The document outlines objectives to study 1D and 2D discrete Fourier transforms and their properties. It also describes various image transforms including the Fourier, Walsh, Hadamard, and discrete cosine transforms. Elementary geometric transforms and operations with images like addition, subtraction and multiplication are defined. Finally, the concepts of the two-dimensional discrete Fourier transform and its inverse are explained.

Uploaded by

Adal Arasu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
32 views

Medical Imaging Techniques: Textbook and Materials

This document discusses medical imaging techniques and digital image processing. It provides an overview of textbooks and materials on the topics. The document outlines objectives to study 1D and 2D discrete Fourier transforms and their properties. It also describes various image transforms including the Fourier, Walsh, Hadamard, and discrete cosine transforms. Elementary geometric transforms and operations with images like addition, subtraction and multiplication are defined. Finally, the concepts of the two-dimensional discrete Fourier transform and its inverse are explained.

Uploaded by

Adal Arasu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

1/11/2018

Medical Imaging
1 2
Textbook and Materials
Rafael C. Gonzalez, Richard E. Woods,
Techniques “Digital Image Processing”, 2nd Edition,
Pearson Education, 2003
Digital Image Processing by Jayaraman,
Veerakumar, 2012
Khandpur R.S, Handbook of Biomedical
Instrumentation, 3/e, Tata McGraw
Hill,New Delhi, 2014
Dr. K. Adalarasu
KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

3 4
Reference Textbook and Materials
William K. Pratt, “Digital Image Processing” ,
John Willey ,2001 Rafael C. Gonzalez, Richard E.
Steve Webb, The physics of medical imaging, Woods, “Digital Image Processing”,
Adam Hilger, Bristol, England, Philadelphia,
USA, 1988
2nd Edition, Pearson Education, 2003
Jain A.K., “Fundamentals of Digital Image Power Point Presentation
Processing”, PHI, 1995.

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

1
1/11/2018

5 6

Reference
William K. Pratt, “Digital Image Processing” ,
John Willey ,2001
Millman Sonka, Vaclav Hlavac, Roger Boyle, Digital Image
Broos/Colic, Thompson Learniy, Vision,
“Image Processing Analysis and Machine”, Fundamentals and
1999.
Jain A.K., “Fundamentals of Digital Image Transforms
Processing”, PHI, 1995.
Chanda Dutta Magundar, “Digital Image
Processing and Applications”, PHI, 2000
KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

7 8
Operations with Images Geometric Transforms
Addition : p+q
Image addition is used for averaging image to
reduce noise
Subtraction: p-q
Used in medical imaging to remove static
background information
Multiplication : p * q
Used to correct gray level shading resulting from
non uniformities in illumination
Elementary geometric transforms for a planar surface element:
Translation, rotation, dilation, stretching, and shearing

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

2
1/11/2018

9 10

Objective
To study 1-D Fourier transform
To Study 2-D Discrete Fourier
Fourier Transform transform and its properties

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

12
Image Transform Fourier Transform
Fourier Transform and DFT Let f(x) be a continuous function of real
Walsh Transform variable x then
Hadamard Transform
Walsh-Hadamard Transform (WHT)
Discrete Cosine Transform (DCT)
Haar Transform
Slant Transform Fourier transform pair exists if f(x) is
Karhunen – Loeve continuous and integrable and F(ω) is
integrable

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

3
1/11/2018

13 14
Discrete Fourier Transform
Discrete Fourier Transform One-Dimensional Fourier Transform and its
Inverse
Discrete form

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

15
Two-Dimensional DFT and Its 16

Discrete Fourier Transform Inverse


Discrete Fourier transform of a function
Relationship is useful when measurements (image) f(x, y) of size M X N is given by the
are an issue in images being processed equation

Variables u and v are transform or


frequency variables
X and y are spatial or image variables

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

4
1/11/2018

Two-Dimensional DFT and Its 17


Two-Dimensional DFT and Its 18

Inverse Inverse
IFT Origin of Fourier transform located at u =
M/2 and v = N/2
When implementing Fourier transform in a
computer actual center of transform
u = (M/2) + 1 and v = (N/2) + 1

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

Two-Dimensional DFT and Its 19


2D - DFT and Its Inverse
20

Inverse
Value of Fourier transform at origin is
“equal to average gray level of image”
White rectangle of size 20 X 40 pixels
superimposed on a black background of
size 512 X 512 pixels
Image was multiplied by (-1)x+y prior to
computing Fourier transform
Order to center spectrum

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

5
1/11/2018

21 22

2D - DFT and Its Inverse Basic Steps for Filtering in


Frequency Domain
Log transformation
To enhance gray-level detail
Value of c = 0.5 was used in
transformation in order to decrease
overall intensity

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

23 24

2D Image Transformation
Translation

Scaling Properties of 2-D


Fourier Transform
Rotation

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

6
1/11/2018

25 26

Properties of 2-D Fourier Transform


Translation

Similarly

When u0 = M/2 and v0 = N/2, it follows

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

27 28

Shifting (Translation) Distributivity and Scaling


From definition of continuous and discrete
Fourier transform pair

In general

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

7
1/11/2018

29 30

Distributivity and scaling Rotation


Introduce polar coordinates
Fourier transform and its inverse are
distributive over addition but not over
multiplication
For two scalars a and b,

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

31 32

Rotation Periodicity & Conjugate Symmetry


Expression indicates that rotating Discrete Fourier transform and its inverse are
periodic with period N
f(x, y) by an angle Ɵ0 rotates F(u, v)
F(u, v)=F(u+N, v)=F(u, v+N)=F(u+N, v+N)
by same angle
Conjugate symmetry
Similarly, rotating F(u, v) rotates f(x,
y) by same angle
Spectrum also is symmetric about origin

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

8
1/11/2018

33 34
Separability
Discrete Fourier transform can be expressed Computation of 2-D Fourier transform as a
in separable form series of 1-D transforms

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

35 36

Separability Questions
Define Discrete Fourier Transform?
What are the properties of Fourier
Transform?

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

9
1/11/2018

37 38

Fast Fourier Transform (FFT) Fast Fourier Transform (FFT)


Computing 1-D Fourier transform of M
points directly requires on order of M2 FFT algorithm developed based on the so-
multiplication/addition operations called successive doubling method
FFT accomplishes same task on order of M
log2 M operations
Example
M = 1024, brute-force method will require
approximately 106 operations
While FFT will require approximately 104
operations

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

39 40

Fast Fourier Transform (FFT) Fast Fourier Transform (FFT)


Computing 1-D Fourier transform of M points FFT algorithm developed based on so-called
directly requires on order of M2 “successive doubling method”
multiplication/addition operations
FFT accomplishes same task on order of M
log2 M operations
Example
M = 1024, brute-force method will require
approximately 106 operations
While FFT will require approximately 104
operations

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

10
1/11/2018

41 42
Fast Fourier Transform (FFT) Fast Fourier Transform (FFT)
M = 2n
n being a positive integer
M = 2k
Where k - positive integer

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

43 44
Fast Fourier Transform (FFT) Fast Fourier Transform (FFT)
Suppose that n = 1 If n=2
Four-point transform can be divided into two
Two-point transform requires parts
evaluation of F(0); then F(1) Positive integer value of n leads to recursive
F(0) first requires computing expressions for number of multiplications and
additions required to implement FFT
Feven(0) and Fodd(O)
One multiplication of Fodd(0) by W 02
and one addition yield F(0)

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

11
1/11/2018

45 46

Fast Fourier Transform (FFT) Fast Fourier Transform (FFT)


For instance, when n = 15 (32,768 points)
FFT has nearly a 2200 to 1 advantage over
the DFT

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

47 48

Comparison Of Various Transforms Comparison Of Various Transforms

KA – MIT – Unit I – Dec, 2017, Sastra University KA – MIT – Unit I – Dec, 2017, Sastra University

12

You might also like