Frequency domain analysis and

Fourier Transform

xkcd.com
 How to Represent Signals?
• Option 1: Taylor series represents any function using
polynomials.

• Polynomials are not the best - unstable and not very

physically meaningful.

• Easier to talk about “signals” in terms of its “frequencies”

(how fast/often signals change, etc).
 Jean-Baptiste Joseph Fourier (1768-1830)
• The big idea (1807):
• Any periodic function can be rewritten as a weighted sum of
Sines and Cosines of different frequencies.
– Not translated into English until 1878!
– called Fourier Series
– Possibly the greatest tool used in Engineering
 A Sum of Sinusoids
• Our building block:

Asin(x +  )
• Add enough of them to get
any signal f(x) you want!

• How many degrees of

freedom?

• What does each control?

• Which one encodes the

coarse vs. fine structure of
the signal?
 Fourier Transform
• We want to understand the frequency  of our signal. So, let’s
reparametrize the signal by  instead of x:

Fourier
f(x) F()
Transform

• For every  from 0 to inf, F() holds the amplitude A and phase  of
the corresponding sine
Asin(x +  )
– How can F hold both? Complex number trick!

F ( ) = R( ) + iI ( )
I ( )
A =  R( ) + I ( )
2 2
 = tan−1

R( )
Inverse Fourier
F() f(x)
Transform
 Time and Frequency

• example : g(t) = sin(2pi f t) + (1/3)sin(2pi (3f) t)

 Time and Frequency

• example : g(t) = sin(2pi f t) + (1/3)sin(2pi (3f) t)

= +
 Frequency Spectra

• example : g(t) = sin(2pi f t) + (1/3)sin(2pi (3f) t)

= +
 Frequency Spectra

• Usually, frequency is more interesting than the phase

Frequency Spectra

= +

=
Frequency Spectra

= +

=
Frequency Spectra

= +

=
Frequency Spectra

= +

=
Frequency Spectra

= +

=
Frequency Spectra


1
= A sin(2 kt )
k =1 k
Frequency Spectra
 FT: Just a change of basis

M * f(x) = F()

* =

.
.
.
 IFT: Just a change of basis

M-1 * F() = f(x)

* =

.
.
.
 Fourier Transform – more formally
Represent the signal as an infinite weighted sum
of an infinite number of sinusoids

F (u ) =  f (x )e −i 2ux
dx
−

Note: e = cos k + i sin k i = −1

ik

Arbitrary function Single Analytic Expression

Spatial Domain (x) Frequency Domain (u)
(Frequency Spectrum F(u))

Inverse Fourier Transform (IFT)


f (x ) =  F (u )e i 2ux
dx
−
 Fourier Transform

• Also, defined as:


F (u ) =  f (x )e −iux
dx
−

Note: e ik = cos k + i sin k i = −1

• Inverse Fourier Transform (IFT)


f (x ) = F (u )eiux dx
1
2 
−
 Fourier Transform and Convolution
Let g = f h

Then G(u ) =  g (x )e −i 2ux dx
−

 
=  f ( )h(x −  )e −i 2ux ddx
− −

  f ( )e  
 
= −i 2u
d h( x −  )e −i 2u ( x − )dx
− −

=
−

 f ( )e −i 2u
d  h(x')e


−
−i 2ux '

dx'

= F (u )H (u )

Convolution in spatial domain

 Multiplication in frequency domain
 Fourier Transform and Convolution

Spatial Domain (x) Frequency Domain (u)

g = f h G = FH
g = fh G = F H

So, we can find g(x) by Fourier transform

g = f  h
IFT FT FT

G = F  H
 Properties of Fourier Transform
Spatial Domain (x) Frequency Domain (u)

Linearity c1 f ( x ) + c2 g ( x ) c1 F (u ) + c2G (u )
1 u
Scaling f (ax ) F 
a a
Shifting f ( x − x0 ) e − i 2ux0 F (u )
Symmetry F (x ) f (− u )
Conjugation f  (x ) F  (− u )
Convolution f (x )  g (x ) F (u )G (u )
d n f (x )
Differentiation (i 2u )n F (u )
dx n
Note that these are derived using
frequency ( e − i 2ux)
 Example use: Smoothing/Blurring
• We want a smoothed function of f(x)

g ( x ) = f ( x )  h( x )

• Let us use a Gaussian kernel h( x )

 1 x2 
h( x ) =
1
exp− 2 
2   2  
x
• Then H (u )
 1 2 2
1
H (u ) = exp− (2u )   2
 2 
G (u ) = F (u )H (u ) u

H(u) attenuates high frequencies in F(u) (Low-pass Filter)!

 2i ( ux + vy )
The 2D Basis Functions e
V

u=-2, v=2 u=-1, v=2 u=0, v=2 u=1, v=2 u=2, v=2

u=-2, v=1 u=-1, v=1 u=0, v=1 u=1, v=1 u=2, v=1

u=-2, v=0 u=-1, v=0 u=0, v=0 u=1, v=0 u=2, v=0

u=-2, v=-1 u=-1, v=-1 u=0, v=-1 u=1, v=-1 u=2, v=-1

u=-2, v=-2 u=-1, v=-2 u=0, v=-2 u=1, v=-2 u=2, v=-2

The wavelength is
1 / u 2 + v. 2 The direction is u/v .
 Frequency Bands
Image Fourier Spectrum

Percentage of image power enclosed in circles (small to large) :

90%, 95%, 98%, 99%, 99.5%, 99.9%

 High Frequency Emphasis
Original High pass Filter

High Frequency High Frequency Emphasis

Emphasis +
Histogram Equalization
 Rotation
2D Image 2D Image - Rotated

Fourier Spectrum Fourier Spectrum

Fourier Transform -- Examples
Image Frequency
Domain Domain
Fourier Transform -- Examples
Image
Fourier spectrum
Image Processing in the Fourier Domain
Magnitude of the FT

Does not look anything like what we have seen

Image Processing in the Fourier Domain

Magnitude of the FT

Does not look anything like what we have seen

 1 0 -1
2 0 -2
Filtering in spatial domain
1 0 -1

* =
Filtering in frequency domain
FFT

FFT

Inverse FFT
=
Slide: Hoiem
 Low-pass Filtering

Let the low frequencies pass and eliminating the high frequencies.

Generates image with overall

shading, but not much detail
 High-pass Filtering

Lets through the high frequencies (the detail), but eliminates the low
frequencies (the overall shape). It acts like an edge enhancer.
Most information at low frequencies!
Fun with Fourier Spectra