Filtering in the Frequency Domain

Outline

► Fourier Transform

► Filtering in Fourier Transform Domain

1/22/2025 2
 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

1/22/2025 3
 Preliminary Concepts

1/22/2025 4
 Fourier Series

1/22/2025 5
 Impulses and the Sifting Property (1)

1/22/2025 6
 Impulses and the Sifting Property (2)

1/22/2025 7
 Impulses and the Sifting Property (3)

1/22/2025 8
 Fourier Transform: One Continuous Variable

1/22/2025 9
 Fourier Transform: One Continuous Variable

1/22/2025 10
 Fourier Transform: One Continuous Variable

1/22/2025 11
 Fourier Transform: Impulses

1/22/2025 12
 Fourier Transform: Impulse Trains

1/22/2025 13
 Fourier Transform: Impulse Trains

1/22/2025 14
 Fourier Transform: Impulse Trains

1/22/2025 15
 Fourier Transform and Convolution

1/22/2025 16
 Fourier Transform and Convolution

Fourier Transform Pairs

1/22/2025 17
 Fourier Transform of Sampled Functions

1/22/2025 18
 Fourier Transform of Sampled Functions

1/22/2025 19
 Question
The Fourier transform of the
sampled function (shown in
the
following figure) is

1. Continuous

2. Discrete

1/22/2025 20
 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,

1/22/2025 21
 Fourier Transform of Sampled Functions

1/22/2025 22
 Nyquist–Shannon sampling theorem

► We can recover from its sampled version if we can

isolate a copy of from the periodic sequence of copies
of this function contained in , the transform of the
sampled function

► Sufficient separation is guaranteed if

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
1/22/2025 23
 Nyquist–Shannon sampling theorem

1/22/2025 24
 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 .

1/22/2025 25
 Aliasing

1/22/2025 26
 Aliasing

1/22/2025 27
 Function Reconstruction from Sampled Data

1/22/2025 28
 The Discrete Fourier Transform (DFT) of One
Variable

1/22/2025 29
 2-D Impulse and Sifting Property: Continuous

1/22/2025 30
 2-D Impulse and Sifting Property: Discrete

1/22/2025 31
 2-D Fourier Transform: Continuous

1/22/2025 32
 2-D Fourier Transform: Continuous

1/22/2025 33
 2-D Sampling and 2-D Sampling Theorem

1/22/2025 34
 2-D Sampling and 2-D Sampling Theorem

1/22/2025 35
 2-D Sampling and 2-D Sampling Theorem

1/22/2025 36
 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

1/22/2025 37
 Aliasing in Images: Example

Re-sampling

1/22/2025 38
 Aliasing in Images: Example

Re-sampling

1/22/2025 39
 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

1/22/2025 40
 2-D Discrete Fourier Transform and Its
Inverse

1/22/2025 41
 Properties of the 2-D DFT
relationships between spatial and frequency intervals

1/22/2025 42
 Properties of the 2-D DFT
translation and rotation

1/22/2025 43
 Properties of the 2-D DFT
periodicity

1/22/2025 44
 Properties of the 2-D DFT
periodicity

1/22/2025 45
 Properties of the 2-D DFT
Symmetry

1/22/2025 46
 Properties of the 2-D DFT
Fourier Spectrum and Phase Angle

1/22/2025 47
1/22/2025 48
1/22/2025 49
 Example: Phase Angles

1/22/2025 50
Example: Phase Angles and The Reconstructed

1/22/2025 51
 The Basic Filtering in the Frequency Domain

Why is the spectrum at

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

1/22/2025 52
 The Basic Filtering in the Frequency Domain

► Modifying the Fourier transform of an image

► Computing the inverse transform to obtain the processed

result

1/22/2025 53
 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?

1/22/2025 54
 The Basic Filtering in the Frequency Domain

1/22/2025 55
 Zero-Phase-Shift Filters

Filters affect the real and imaginary parts equally,

and thus no effect on the phase.
These filters are called zero-phase-shift filters

1/22/2025 56
 Examples: Nonzero-Phase-Shift Filters

Even small
Phasechanges
angle is in the phase angle can ishave
Phase angle
dramaticmultiplied
(usually by undesirable) effects on the
multiplied by filtered
output 0.5 0.5

1/22/2025 57
 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*, H(u,v), of

size PxQ with center at coordinates (P/2, Q/2)
*generate from a given spatial filter, we pad the spatial filter, multiply the expadded
array by (-1)x+y, and compute the DFT of the result to obtain a centered H(u,v).
1/22/2025 58
 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

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

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

1/22/2025 59
 An Example:
Steps for Filtering in the Frequency Domain

1/22/2025 60
 Correspondence Between Filtering in the
Spatial and Frequency Domains (1)

1. Both components are Gaussian and real

2. The functions behave reciprocally

1/22/2025 61
 Correspondence Between Filtering in the
Spatial and Frequency Domains (2)

1/22/2025 62
 Correspondence Between Filtering in the
Spatial and Frequency Domains (3)

1/22/2025 63
 Correspondence Between Filtering in the
Spatial and Frequency Domains: Example

600x600

1/22/2025 64
 Correspondence Between Filtering in the
Spatial and Frequency Domains: Example

1/22/2025 65
 Generate H(u,v)

1/22/2025 66
 Generate H(u,v)

1/22/2025 67
 Image Smoothing Using Filter Domain Filters:
ILPF

1/22/2025 68
 ILPF Filtering
Example

Ringing artifacts

1/22/2025 69
 Image Smoothing Using Filter Domain Filters:
ILPF

1/22/2025 70
 ILPF Filtering Example

1/22/2025 71
 The Spatial Representation of ILPF

1/22/2025 72
 Image Smoothing Using Filter Domain Filters:
BLPF

1/22/2025 73
1/22/2025 74
 The Spatial Representation of BLPF

1/22/2025 75
 Image Smoothing Using Filter Domain Filters:
GLPF

1/22/2025 76
 Image Smoothing Using Filter Domain Filters:
GLPF

1/22/2025 77
1/22/2025 78
 Examples of smoothing by GLPF (1)

1/22/2025 79
 Examples of smoothing by GLPF (2)

1/22/2025 80
 Examples of smoothing by GLPF (3)

1/22/2025 81
Question

82
 Question

In Figure below, why multiplying the phase angle by −1 flipped

the image with respect to both coordinate axes ?

83
 Image Sharpening Using Frequency Domain
Filters

A highpass filter is obtained from a given lowpass filter

using

1/22/2025 84
 Image Sharpening Using Frequency Domain
Filters

1/22/2025 85
1/22/2025 86
 Filtering Results by IHPF

1/22/2025 87
 Filtering Results by BHPF

1/22/2025 88
 Filtering Results by GHPF

1/22/2025 89
 Using Highpass Filtering and Threshold for
Image Enhancement

BHPF
(order 4 with a cutoff
frequency 50)

1/22/2025 90
 The Laplacian in the Frequency Domain

1/22/2025 91
 The Laplacian in the Frequency Domain

1/22/2025 92
 The Laplacian in the Frequency Domain

1/22/2025 93
 Unsharp Masking, Highboost Filtering and
High-Frequency-Emphasis Fitering

1/22/2025 94
 Unsharp Masking, Highboost Filtering and
High-Frequency-Emphasis Fitering

1/22/2025 95
 Gaussian Filter
D0=40

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

1/22/2025 96
 Homomorphic Filtering

= ?

1/22/2025 97
 Homomorphic Filtering

1/22/2025 98
 Homomorphic Filtering

The illumination component of an image generally is

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

1/22/2025 99
 Homomorphic Filtering

Attenuate the contribution

made by illumination and
amplify the contribution made
by reflectance

1/22/2025 100
 Homomorphic Filtering

1/22/2025 101
 Homomorphic Filtering

1/22/2025 102
 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

1/22/2025 103
 Selective Filtering:
Bandreject and Bandpass Filters

1/22/2025 104
 Selective Filtering:
Bandreject and Bandpass Filters

1/22/2025 105
 Example: Image Denoising by Using DCT Transform

Weeks 1 & 2 106

 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.

1/22/2025 107
 Selective Filtering:
Notch Filters

1/22/2025 108
 Examples:
Notch Filters
(1)

1/22/2025 109
 Examples:
Notch Filters
(2)

1/22/2025 110
1/22/2025 111
Question

112