CS589-04 Digital Image Processing

Lecture 3. Filtering in the

Frequency Domain

Spring 2008
New Mexico Tech
 Outline

► Fourier Transform

► Filtering in Fourier Transform Domain

11/08/24 2
 Fourier Series and Fourier Transform:
History
► Jean Baptiste Joseph Fourier, French mathematician and
physicist (03/21/1768-05/16/1830)
http://en.wikipedia.org/wiki/Joseph_Fourier
Permanent
Orphaned: at nine
Secretary of the
French Academy
Egyptian of Sciences:
expedition with 1822
Napoleon I: Théorie
1798 analytique de la
Governor of chaleur : 1822
Lower Egypt
(The Analytic
Theory of Heat)

11/08/24 3
 Fourier Series and Fourier Transform:
History
► Fourier Series
Any periodic function can be expressed as the sum
of sines and /or cosines of different frequencies,
each multiplied by a different coefficients

► Fourier Transform
Any function that is not periodic can be expressed
as the integral of sines and /or cosines multiplied
by a weighing function

11/08/24 4
 Fourier Series: Example

11/08/24 5
 Preliminary Concepts

j   1, a complex number
C R  jI
the conjugate
C* R - jI

| C | R 2  I 2 and  arctan( I / R)
C | C | (cos   j sin  )
Using Euler's formula,
C | C | e j

11/08/24 6
 Fourier Series

A function f (t ) of a continuous variable t that is periodic

with period, T , can be expressed as the sum of sines and
cosines multiplied by appropriate coefficients
 2 n
j t
f (t )   cn e T

n  

where
2 n
1 T /2  j t
cn   f (t )e T
dt for n 0, 1, 2,...
T  T /2

11/08/24 7
 Impulses and the Sifting Property (1)
A unit impulse of a continuous variable t located
at t =0, denoted  (t ), defined as
 if t 0
 (t ) 
0 if t 0
and is constrained also to satisfy the identity

  (t )dt 1



The sifting property  f (t ) (t  t0 )dt  f (t0 )




f (t ) (t )dt  f (0)
11/08/24 8
 Impulses and the Sifting Property (2)
A unit impulse of a discrete variable x located
at x =0, denoted  ( x), defined as
1 if x 0
 ( x) 
0 if x 0
and is constrained also to satisfy the identity


  ( x) 1
x  

The sifting property 
x  
f ( x) ( x  x0 )  f ( x0 )



x  
f ( x) ( x)  f (0)
11/08/24 9
 Impulses and the Sifting Property (3)

impulse train sT (t ),


sT (t )    (t  nT )
n  

11/08/24 10
 Fourier Transform: One Continuous
Variable

The Fourier Transform of a continous function f (t )


F (  ) { f (t )}  f (t )e  j 2t dt


The Inverse Fourier Transform of F (  )


f (t )  {F (  )}  F (  )e j 2t d 
1


11/08/24 11
 Fourier Transform: One Continuous
Variable

 W /2
F (  )  f (t )e  j 2 t
dt  Ae  j 2t dt
  W /2

A  j 2 t W /2 A
  e    e jW  e  jW 
j 2  W /2 j 2 W
sin(W )
 AW
(W )

11/08/24 12
 Fourier Transform: Impulses

The Fourier transform of a unit impulse located at the origin:


F (  )   (t )e  j 2t dt


e  j 2 0
=1
The Fourier transform of a unit impulse located at t t0 :

F (  )   (t  t0 )e  j 2t dt


e  j 2t0
=cos(2 t0 )  j sin (2 t0 )

11/08/24 13
 Fourier Transform: Impulse Trains

Impulse train sT (t ), sT (t )    (t  nT )
n  

The Fourier series:

 2 n
j t
sT (t )   c n e T

n  

where
2 n
1 T /2  j t
cn 
T   T /2
sT (t )e T
dt

11/08/24 14
 Fourier Transform: Impulse Trains

 j 21Tn t T /2  j 2Tjn2tTn t j 2t 1 T /2 j 2T nj t2Tnt j 2t

cn e 

T 

 

T /2 
sT (et )e e dt = dt 
 T 

 T /2

e(t )e e dt dt
1 0 1
= e  n
 Tj 2 (   T) t n
 e T
dt  (   )
 T
 2 n n  2 n n

  ( j T t ) 1  ( j  t )e j 2t du

1 
sT (t )  c n e T    e T T
n  2 nt T n  
j
e T

11/08/24 15
 Fourier Transform: Impulse Trains
Let S (  ) denote the Fourier transform of the
periodic impulse train S T (t )
2 n
 1  j t 
S (  )  S T (t )   e T

 T n   
1   j 2Tn t 
   e 
T  n   
1  n
= 
T n  
 ( 
T
)

11/08/24 16
Fourier Transform and Convolution
The convolution of two functions is denoted
by the operator

f (t ) h(t )  f ( )h(t   )d

 
 f (t ) h(t )   f ( )h(t   )d  e  j 2t dt

     
 
=  f ( )  h(t   )e  j 2t dt  d

    

=  f ( )  H (  )e  j 2  d


=H (  )  f ( )e  j 2 d


=H (  ) F (  )
11/08/24 17
Fourier Transform and Convolution

Fourier Transform Pairs

f (t ) h(t )  H (  ) F (  )

f (t )h(t )  H (  ) F (  )

11/08/24 18
 Fourier Transform of Sampled
Functions

f (t )  f (t ) sT (t )

  f (t ) (t  nT )
n  

11/08/24 19
 Fourier Transform of Sampled
Functions

 
F (  )  f (t )  f (t ) sT (t ) F (  ) ? S (  )

F (  ) F1(  ) S (  )  n  F ( ) S (    )d
S ( )  
1T n  
 ( 
 T
)
n
=  F ( )   (     )d
T  
n   T
1   n
= 
T n     
F ( ) (    
T
) d

1  n
= 
T n  
F ( 
T
)

11/08/24 20
 Question

The Fourier transform of the

sampled function (shown in the
following figure) is

1. Continuous

2. Discrete

11/08/24 21
 Fourier Transform of Sampled
Functions
► A bandlimited signal is a signal whose Fourier
transform is zero above a certain finite frequency.
In other words, if the Fourier transform has finite
support then the signal is said to be bandlimited.

An example of a simple bandlimited signal is a

sinusoid of the form,

x(t ) sin(2 ft   )

11/08/24 22
 Fourier Transform of Sampled
Functions

 max max
Over-sampling
1
 2 max
F (  )  T

1 n Critically-sampling
T
 F ( 
T
)
n   1
2 max
T
under-sampling
1
 2 max
11/08/24 T 23
 Nyquist–Shannon sampling theorem

► We can recoverf (t ) from its sampled version if we

can isolate a copy
F ( of
) from the periodic sequence
F (  )
of copies of this function contained in , the
transform of the fsampled
(t ) function

►
1
Sufficient separation is guaranteed if  2  max
T
Sampling theorem: A continuous, band-limited
function can be recovered completely from a set of
its samples if the samples are acquired at a rate
exceeding twice the highest frequency content of the
function
11/08/24 24
 Nyquist–Shannon sampling theorem

?

f (t )  F (  )e j 2t d 


11/08/24 25
 Aliasing

If a band-limited function is sampled at a rate that is

less than twice its highest frequency?

The inverse transform will yield a corrupted

function. This effect is known as frequency
aliasing or simply as aliasing.

11/08/24 26
 Aliasing

11/08/24 27
 Aliasing

11/08/24 28
Function Reconstruction from Sampled
Data

F (  ) H (  ) F (  )

f (t )  1
 F (  )
 1
 H ( ) F ( )
h(t ) f (t )

f (t )   f (nT )sinc  (t  nT ) / nT 
n  
11/08/24 29
The Discrete Fourier Transform (DFT) of
One Variable

M1
F (  )   f ( x)e  j 2 x / M ,  0,1,..., M  1
x 0

M1
1
f ( x) 
M
 F
 0
(  ) e j 2 x / M
, x 0,1, 2,..., M  1

11/08/24 30
 2-D Impulse and Sifting Property:
Continuous
 if t  z 0
The impulse  (t , z ),  (t , z ) 
0 otherwise
 
and    (t , z )dtdz 1
 

The sifting property

 

 
f (t , z ) (t , z )dtdz  f (0, 0)
and
 

 
f (t , z ) (t  t0 , z  z0 )dtdz  f (t0 , z0 )

11/08/24 31
 2-D Impulse and Sifting Property:
Discrete

1 if x  y 0
The impulse  ( x, y ),  ( x, y ) 
0 otherwise

The sifting property

 


x   y  
f ( x, y ) ( x, y )  f (0, 0)

and
 


x   y  
f ( x, y ) ( x  x0 , y  y0 )  f ( x0 , y0 )

11/08/24 32
 2-D Fourier Transform: Continuous

 
F (  , )   f (t , z )e  j 2 (  t  z )
dtdz
 

and
 
f (t , z )   f (  , )e j 2 (  t  z )
d  d
 

11/08/24 33
 2-D Fourier Transform: Continuous
 
F (  , ) 
 f (t , z )e  j 2 ( t  z ) dtdz
 
T /2 Z /2
  Ae  j 2 ( t  z ) dtdz
 T /2  Z /2

 sin(T )   sin( T ) 
 ATZ     T 
  T  

11/08/24 34
 2-D Sampling and 2-D Sampling
Theorem
2  D impulse train:
 
sT Z (t , z )     (t  mT , z  nZ )
m   n  

11/08/24 35
 2-D Sampling and 2-D Sampling
Theorem
Function f (t , z ) is said to be band-limited if its Fourier transform
is 0 outside a rectangle established by the intervals [-max ,max ]
and [- max , max ], that is
F (  , ) 0 for |  | max and |  | max

Two-dimensional sampling theorem:

A continuous, band-limited function f (t , z ) can be recovered with
no error from a set of its samples if the sampling intervals are
1 1
T< and Z<
2 max 2 max
11/08/24 36
 2-D Sampling and 2-D Sampling
Theorem

11/08/24 37
 Aliasing in Images: Example
In an image system, the
number of samples is fixed
at 96x96 pixels. If we use
this system to digitize
checkerboard patterns …

Under-sampling

11/08/24 38
 Aliasing in Images: Example

Re-sampling

11/08/24 39
 Aliasing in Images: Example

Re-sampling

11/08/24 40
 Moiré patterns

► Moiré patterns are often an undesired artifact of

images produced by various digital imaging and
computer graphics techniques
e. g., when scanning a halftone picture or ray
tracing a checkered plane. This cause of moiré is a
special case of aliasing, due to under-sampling a
fine regular pattern

http://en.wikipedia.org/wiki/Moiré_pattern

11/08/24 41
 Moiré patterns

11/08/24 42
 Moiré patterns

11/08/24 43
 Moiré patterns
A moiré pattern
formed by
incorrectly
down-sampling
the former
image

11/08/24 44
2-D Discrete Fourier Transform and
Its Inverse
DFT:
M 1 N1  j 2 (  x / M  y / N )

F (  , )    f ( x, y )e
x 0 y 0

 0,1, 2,..., M  1; 0,1, 2,..., N  1;

f ( x, y ) is a digital image of size M N.

IDFT:
M 1 N1 j 2 (  x / M  y / N )
1
f ( x, y ) 
MN
  F ( , )e
x 0 y 0

11/08/24 45
 Properties of the 2-D DFT
relationships between spatial and frequency
intervals

Let T and Z denote the separations between samples,

then the seperations between the corresponding discrete,
frequency domain variables are given by
1
 
M T
1
and  
N Z

11/08/24 46
 Properties of the 2-D DFT
translation and rotation

f ( x, y )e j 2 ( 0 x / M  0 y / N )  F (   0 ,   0 )
and
 j 2 (  x0 / M  y0 / N )
f ( x - x0 , y - y0 )  F (  , )e

Using the polar coordinates

x r cos  y=rsin  =cos  = sin 
results in the following transform pair:
f (r ,    0 )  F ( ,    0 )
11/08/24 47
 Properties of the 2-D DFT
periodicity

2  D Fourier transform and its inverse are infinitely periodic

F (  , ) F (   k1M , ) F (  ,  k 2 N ) F (   k1M ,  k 2 N )

f ( x, y )  f ( x  k1M , y )  f ( x, y  k2 N )  f ( x  k1M , y  k2 N )

f ( x ) e j 2 (  0 x / M )  F (    0 )
0 M / 2, f ( x)(  1) x  F (   M / 2)

f ( x, y )( 1) x  y  F (   M / 2,  N / 2)

11/08/24 48
 Properties of the 2-D DFT
periodicity

11/08/24 49
 Properties of the 2-D DFT
Symmetry

11/08/24 50
 Properties of the 2-D DFT
Fourier Spectrum and Phase Angle

2-D DFT in polar form

F (u, v) | F (u, v) | e j (u ,v )
Fourier spectrum
1/2
| F (u, v) | R (u, v)  I (u, v) 
2 2

Power spectrum
P(u, v) | F (u, v) |2 R 2 (u, v)  I 2 (u, v)
Phase angle
 I (u , v) 
 (u,v)=arctan  
 R (u , v ) 

11/08/24 51
11/08/24 52
11/08/24 53
 Example: Phase Angles

11/08/24 54
xample: Phase Angles and The Reconstructe

11/08/24 55
 2-D Convolution Theorem
1-D convolution
M1
f ( x ) h ( x )   f ( m )h ( x  m )
m 0

2-D convolution
M 1 N1
f ( x , y ) h ( x , y )    f ( m, n ) h ( x  m , y  n )
m 0 n 0

x 0,1, 2,..., M  1; y 0,1, 2,..., N  1.

f ( x , y ) h ( x , y )  F ( u, v ) H ( u, v )
f ( x , y )h ( x , y )  F ( u, v ) H ( u, v )
11/08/24 56
 An Example of Convolution

Mirroring h
about the
origin

Translating
the mirrored
function by
x

Computing
the sum for
each x
11/08/24 57
 An Example of Convolution

It causes the
wraparou
nd error

It can be
solved by
appending
zeros

11/08/24 58
 Zero Padding

► Consider two functions f(x) and h(x) composed of A

and B samples, respectively

► Append zeros to both functions so that they have

the same length, denoted by P, then wraparound is
avoided by choosing

P ≥A+B-1

11/08/24 59
 Zero Padding

► Let f(x,y) and h(x,y) be two image arrays of sizes

A×B and C×D pixels, respectively. Wraparound
error in their convolution can be avoided by
padding these functions with zeros

 f ( x, y ) 0  x  A -1 and 0  y  B -1
f p ( x, y ) 
 0 A  x  P or B  y Q

h( x, y ) 0  x C -1 and 0  y  D -1
h p ( x, y ) 
 0 C  x  P or D  y Q

Here P  A  C  1; Q  B  D  1
11/08/24 60
 Summary

11/08/24 61
 Summary

11/08/24 62
 Summary

11/08/24 63
 Summary

11/08/24 64
 The Basic Filtering in the Frequency
Domain

Why is the spectrum at

almost ±45 degree
stronger than the
spectrum at other
directions?

11/08/24 65
 The Basic Filtering in the Frequency
Domain
► Modifying the Fourier transform of an image

► Computing the inverse transform to obtain the

processed result

g ( x, y )  1{H (u , v) F (u , v)}

F (u , v) is the DFT of the input image

H (u , v) is a filter function.

11/08/24 66
 The Basic Filtering in the Frequency
Domain
► In a filter H(u,v) that is 0 at the center of the
transform and 1 elsewhere, what’s the output
image?

11/08/24 67
 The Basic Filtering in the Frequency
Domain

11/08/24 68
 The Basic Filtering in the Frequency
Domain

11/08/24 69
 Zero-Phase-Shift Filters
1
g ( x, y )  {H (u, v) F (u, v)}
F (u , v) R (u , v)  jI (u , v)
1
g ( x, y )   H (u, v) R(u, v)  jH (u, v) I (u, v) 
Filters affect the real and imaginary parts equally,
and thus no effect on the phase.
These filters are called zero-phase-shift filters

11/08/24 70
 Examples: Nonzero-Phase-Shift
Filters

Even small changes in the phase

Phase angle Phase angle
angle can have
dramatic (usually undesirable)is effects
is multiplied multipliedon the
by 0.5 by 0.5
filtered output
11/08/24 71
 Summary:
Steps for Filtering in the Frequency
Domain
1. Given an input image f(x,y) of size MxN, obtain
the padding parameters P and Q. Typically, P = 2M
and Q = 2N.

2. Form a padded image, fp(x,y) of size PxQ by

appending the necessary number of zeros to
f(x,y)

3. Multiply fp(x,y) by (-1)x+y to center its transform

4. Compute the DFT, F(u,v) of the image from

step 3

5. Generate a real, symmetric filter function*,

11/08/24 72

H(u,v), of size PxQ with center at coordinates

 Summary:
Steps for Filtering in the Frequency
Domain
6. Form the product G(u,v) = H(u,v)F(u,v) using
array multiplication

7. Obtain the processed image

 
g p ( x, y )  real   1  G (u , v )   (  1) x  y

8. Obtain the final processed result, g(x,y), by

extracting the MxN region from the top, left
quadrant of gp(x,y)

11/08/24 73
 An Example:
Steps for Filtering in the Frequency
Domain

11/08/24 74
Correspondence Between Filtering in the
Spatial and Frequency Domains (1)

Let H(u) denote the 1-D frequency domain Gaussian filter

- u 2 /2 2
H (u )  Ae

The corresponding filter in the spatial domain

 2 2 2 x 2
h( x)  2 Ae

1. Both components are Gaussian

and real
2. The functions behave reciprocally
11/08/24 75
Correspondence Between Filtering in the
Spatial and Frequency Domains (2)
Let H (u ) denote the difference of Gaussian filter
- u 2 /212 - u 2 /2 22
H (u )  Ae  Be
with A B and  1  2

The corresponding filter in the spatial domain

 2 212 x 2  2 2 22 x 2
h( x)  2 1 Ae  2 2 Ae

High-pass filter or low-pass

filter ?
11/08/24 76
Correspondence Between Filtering in the
Spatial and Frequency Domains (3)

11/08/24 77
Correspondence Between Filtering in the
Spatial and Frequency Domains:
Example

600x600

11/08/24 78
Correspondence Between Filtering in the
Spatial and Frequency Domains:
Example

11/08/24 79
 Generate H(u,v)

 f ( x, y ) 0 x 599 and 0  y 599

f p ( x, y ) 
 0 600  x 602 or 600  y 602

 h ( x, y ) 0  x 2 and 0  y 2
hp ( x, y ) 
0 3  x 602 or 3  y 602

Here P  A(600)  C (3)  1 602;

Q B(600)  D(3)  1 602.

11/08/24 80
 Generate H(u,v)

1. Multiply hp ( x, y ) by (-1) x  y to center the frequency domain filter

2. Compute the forward DFT of the result in (1)

3. Set the real part of the resulting DFT to 0 to account for

parasitic real parts

4. Multiply the result by (-1)u v , which is implicit when h( x, y )

was moved to the center of hp ( x, y ).

11/08/24 81
 Image Smoothing Using Filter Domain
Filters: ILPF

Ideal Lowpass Filters (ILPF)

1 if D(u, v) D0
H (u, v) 
0 if D(u, v)  D0

D0 is a positive constant and D(u, v) is the distance between a point (u, v)

in the frequency domain and the center of the frequency rectangle
2 1/2
D(u, v)  (u  P / 2)  (v  Q / 2) 
2

11/08/24 82
 Image Smoothing Using Filter Domain
Filters: ILPF

11/08/24 83
 ILPF Filtering Example

11/08/24 84
 ILPF
Filtering
Example

11/08/24 85
 The Spatial Representation of ILPF

11/08/24 86
 Image Smoothing Using Filter Domain
Filters: BLPF
Butterworth Lowpass Filters (BLPF) of order n and
with cutoff frequency D0
1
H (u , v) 
1   D(u , v) / D0 
2n

11/08/24 87
11/08/24 88
 The Spatial Representation of BLPF

11/08/24 89
 Image Smoothing Using Filter Domain
Filters: GLPF

Gaussian Lowpass Filters (GLPF) in two dimensions is given

 D 2 ( u , v )/2 2
H (u, v) e

By letting  D0
 D 2 ( u , v )/2 D02
H (u, v) e

11/08/24 90
 Image Smoothing Using Filter Domain
Filters: GLPF

11/08/24 91
11/08/24 92
 Examples of smoothing by GLPF (1)

11/08/24 93
 Examples of smoothing by GLPF (2)

11/08/24 94
 Examples of smoothing by GLPF (3)

11/08/24 95
 Image Sharpening Using Frequency
Domain Filters

A highpass filter is obtained from a given lowpass

filter using

H HP (u , v) 1  H LP (u, v)

A 2-D ideal highpass filter (IHPL) is defined as

0 if D(u, v) D0
H (u , v) 
1 if D(u, v)  D0

11/08/24 96
 Image Sharpening Using Frequency
Domain Filters

A 2-D Butterworth highpass filter (BHPL) is defined as

1
H (u , v) 
1   D0 / D(u, v) 
2n

A 2-D Gaussian highpass filter (GHPL) is defined as

 D 2 ( u , v )/2 D02
H (u, v) 1  e

11/08/24 97
11/08/24 98
The Spatial Representation of Highpass
Filters

11/08/24 99
 Filtering Results by IHPF

11/08/24 100
 Filtering Results by BHPF

11/08/24 101
 Filtering Results by GHPF

11/08/24 102
Using Highpass Filtering and Threshold
for Image Enhancement

BHPF
(order 4 with a
cutoff frequency
50)

11/08/24 103
The Laplacian in the Frequency Domain

H (u , v)  4 2 (u 2  v 2 )

H (u , v)  4 2  (u  P / 2) 2  (v  Q / 2) 2 ) 
 4 2 D 2 (u , v)
The Laplacian image
2 f ( x, y )  1  H (u , v) F (u , v)

Enhancement is obtained
g ( x, y )  f ( x, y )  c2 f ( x, y ) c -1
11/08/24 104
The Laplacian in the Frequency Domain

The enhanced image

1
g ( x, y )   F (u, v)  H (u, v) F (u, v)
 1 1  H (u, v)  F (u, v)


 1  1  4 2 D 2 (u, v)  F (u, v)

11/08/24 105
The Laplacian in the Frequency Domain

11/08/24 106
 Unsharp Masking, Highboost Filtering
and High-Frequency-Emphasis Fitering

g mask ( x, y )  f ( x, y )  f LP ( x, y )

f LP ( x, y )  1  H LP (u, v) F (u , v) 

Unsharp masking and highboost filtering

g ( x, y )  f ( x, y )  k * g mask ( x, y )

 
g ( x, y )  1  1  k * 1  H LP (u, v)   F (u, v)
 1 1  k * H HP (u, v)  F (u, v)

11/08/24 107
 Unsharp Masking, Highboost Filtering
and High-Frequency-Emphasis Fitering

1
g ( x, y )   k
1  k2 * H HP (u , v)  F (u , v)
k1 0 and k2 0

11/08/24 108
 Gaussian Filter
D0=40

High-Frequency-Emphasis Filtering
Gaussian Filter
K1=0.5, k2=0.75

11/08/24 109
 Homomorphic Filtering

f ( x, y ) i ( x, y )r ( x, y )
  f ( x, y )  
=  i ( x, y )    r ( x, y )  ?

z ( x, y ) ln f ( x, y ) ln i ( x, y )  ln r ( x, y )

 z ( x, y )  ln f ( x, y )  ln i ( x, y )   ln r ( x, y )

Z (u , v) Fi (u , v)  Fr (u , v)

11/08/24 110
 Homomorphic Filtering

S (u , v) H (u , v) Z (u , v)
H (u, v) Fi (u, v)  H (u, v) Fr (u , v)
s ( x, y )  1  S (u , v)
 1  H (u , v) Fi (u , v)  H (u , v) Fr (u , v)
 1  H (u , v) Fi (u , v)   1  H (u , v) Fr (u , v)
i '( x, y )  r '( x, y )

g ( x, y ) e s ( x , y ) ei '( x , y ) e r '( x , y ) i0 ( x, y )r0 ( x, y )

11/08/24 111
 Homomorphic Filtering

The illumination component of an image generally

is characterized by slow spatial variations, while
the reflectance component tends to vary abruptly

These characteristics lead to associating the low

frequencies of the Fourier transform of the
logarithm of an image with illumination the high
frequencies with reflectance.
11/08/24 112
 Homomorphic Filtering


H (u , v) (H  L ) 1  e
 c  D 2 ( u , v )/ D02 
   
  L

Attenuate the
contribution made by
illumination and amplify
the contribution made by
reflectance

11/08/24 113
 L 0.25
Homomorphic
H 2
Filtering
c 1
D0 80

11/08/24 114
 Homomorphic Filtering

11/08/24 115
 Selective Filtering

Non-Selective Filters:
operate over the entire frequency rectangle

Selective Filters
operate over some part, not entire frequency
rectangle
• bandreject or bandpass: process specific
bands
• notch filters: process small regions of the
frequency rectangle

11/08/24 116
 Selective Filtering:
Bandreject and Bandpass Filters

H BP (u , v) 1  H BR (u , v)

11/08/24 117
 Selective Filtering:
Bandreject and Bandpass Filters

11/08/24 118
 Selective Filtering:
Notch Filters
Zero-phase-shift filters must be symmetric about
the origin.
A notch with center at (u0, v0) must have a
corresponding notch at location (-u0,-v0).

Notch reject filters are constructed as products of

highpass filters whose centers have been translated
to the centers of the notches.
Q
H NR (u, v)  H k (u, v) H  k (u, v)
k 1

where H k (u , v) and H - k (u, v) are highpass filters whose centers are

at (uk , vk ) and (-uk , -vk ), respectively.
11/08/24 119
 Selective Filtering:
Notch Filters
Q
H NR (u , v)  H k (u , v) H  k (u , v)
k 1

where H k (u , v) and H - k (u , v) are highpass filters whose centers are

at (uk , vk ) and (-uk , -vk ), respectively.

A Butterworth notch reject filter of order n


3
1  1 
H NR (u , v)   2n   2n 
k 1  1   D0 k / Dk (u , v )    1   D0 k / D k (u, v)  

2 1/2
Dk (u , v)  (u  M / 2  uk )  (v  N / 2  vk ) 
2

2 1/2
D k (u , v)  (u  M / 2  uk )  (v  N / 2  vk ) 
2

11/08/24 120
Examples:
Notch
Filters (1)

A Butterworth notch
reject filter D 0 =3
and n=4 for all
notch pairs

11/08/24 121
 Examples:
Notch Filters
(2)

11/08/24 122
11/08/24 123