SAMPLING THEORY

Le Thanh Ha, Ph.D

Assoc. Prof. at University of Engineering and Technology,
Vietnam National University
[email protected]; [email protected]; 0983 692 592
 Frequency Spectra
• example : g(t) = sin(2πf t) + (1/3)sin(2π(3f) t)

= +

Slides: Efros
Frequency Spectra
Frequency Spectra

= +

=
Frequency Spectra

= +

=
Frequency Spectra

= +

=
Frequency Spectra

= +

=
Frequency Spectra

= +

=
Frequency Spectra


1
= A sin(2 kt )
k =1 k
 Review: 1D Fourier Transform
A function f(x) can be represented as a weighted combination of phase-
shifted sine waves
+ Inverse Fourier

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

How to compute F(u)?

+

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

−
 Review: 1D Fourier Transform
• Trigonometric identities
eix = cos( x) + i sin( x)
eix + e− ix
cos( x) =
2
eix − e − ix
sin( x) =
2i

• Home work: Calculate Fourier transform of cos(2sx)

 Review: Box Function
f(x)

1 x1
f ( x) =  2
0 x1 x
2

|F(u)|

sin u
F (u ) = = sin c(u )
u u

If f(x) is bounded, F(u) is unbounded

 Review: Dirac Delta and its Transform
f(x)
 x = 0
 ( x) =  +

0 x  0  f ( x) ( x)dx = f (0)

−
+

  ( x)dx = 1
−
x
0

f(x)

−i 2ua
F{ ( x − a)} = e
+∞

න 𝑓 𝑥 𝛿 𝑥 − 𝑎 𝑑𝑥 = 𝑓(𝑎)
−∞

a
 Review: Cosine
1.5

1 
0.5

-0.5 -1 1
-1

-1.5

cos(2x)

If f(x) is even, so is F(u)

 Review: Convolution
• A mathematical operator which computes the “amount of
overlap” between two functions. Can be thought of as a
general moving average

• Continuous domain:

f ( x)  g ( x) =  f (a) g ( x − a )da
 Conv. Example
• Suppose we want to compute the convolution of the following
two functions:

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 16

 Conv. Example
Step 1: find g(-a) Step 2: find g(x-a)

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 17

 Conv. Example
• Step 3: Shift the impulse response function over target
function and take integral at every position
– Case 1 (x<0): no overlap.

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 18

 Conv. Example
• Case 1:

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 19

 Conv. Example
• Case 2 (0 ≤ 𝑥 < 1):

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 20

 Conv. Example

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 21

 Conv. Example
• Case 4: 𝑥 > 2

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 22

 Conv. Example

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 23

 Conv. Example (Wiki)

10/5/2022 Assoc. Prof. Dr. Le Thanh Ha 24

Convolution with Dirac Delta function
Convolution with an “train” of
impulses

=
 Review: Properties
Linearity: af ( x) + bg ( x)  aF (u) + bG(u)

Time shift: f ( x − x0 )  e − i 2ux0 F (u )

df ( x)
Derivative:  uF (u )
x
F (u )
Integration:  f ( x)dx 
u

Convolution: f ( x)  g ( x)  F (u)G(u)
 Sampling Analysis
fs(x)
f(x)

Sampling

x x
… -2T -T 0 T 2T …
 Sampling Analysis
fs(x)
f(x)

Sampling

x Reconstruction x
… -2T -T 0 T 2T …
 Sampling Analysis
fs(x)
f(x)

Sampling

x Reconstruction x
… -2T -T 0 T 2T …

What sampling rate (T) is sufficient to reconstruct the

continuous version of the sampled signal?
 Sampling Theory
• How many samples are required to represent a given signal
without loss of information?

• What signals can be reconstructed without loss for a given

sampling rate?
Sampling Analysis: Spatial Domain
+

f(x)   ( x − kT )
k = −

x … -2T -T 0 T 2T … x

+
f ( x)    ( x − kT )
k = −

?
fs(x)

x
… -2T -T 0 T 2T …
Sampling Analysis: Spatial Domain
+

f(x)   ( x − kT )
k = −

x … -2T -T 0 T 2T … x

+
f ( x)    ( x − kT )
k = −

? What happens in Frequency

fs(x) domain?

x
… -2T -T 0 T 2T …
Fourier Transform of Dirac Comb
 Fourier Transform of Dirac Comb

T 1/T

+ +

  ( x − kT )
k
  (u − )
k = − T
k = −

+

e
k = −
−i 2ukT

Moving the spikes closer together in the spatial domain

moves them farther apart in the frequency domain!
Sampling Analysis: Spatial Domain
+

f(x)   ( x − kT )
k = −

x … -2T -T 0 T 2T … x

+
f ( x)    ( x − kT )
k = −

? What happens in Frequency

fs(x) domain?

x
… -2T -T 0 T 2T …
Sampling Analysis: Freq. Domain
+
n
F(u)   (u − )
n = − T


-fmax fmax u … -1/T 0 1/T … u
Sampling Analysis: Freq. Domain
+
n
F(u)   (u − )
n = − T


-fmax fmax u … -1/T 0 1/T … u

How does the convolution result look like?

Sampling Analysis: Freq. Domain

F(u)  (u)

-fmax fmax u … -1/T 0 1/T … u
Sampling Analysis: Freq. Domain

F(u)  (u)

-fmax fmax u … -1/T 0 1/T … u

F (u)   (u)
Sampling Analysis: Freq. Domain

F(u)  (u)

-fmax fmax u … -1/T 0 1/T … u

F (u)   (u) = G(u)

G(0)?
G(fmax)?
G(u)?
Sampling Analysis: Freq. Domain

F(u)  (u)

-fmax fmax u … -1/T 0 1/T … u

F (u)   (u) = G(u)

G(0) = F(0)
G(fmax) = F(fmax)
G(u) = F(u)
Sampling Analysis: Freq. Domain

F(u)
 (u)
1
How about  (u − )
 T

-fmax fmax u … -1/T 0 1/T … u

F (u)   (u)

Fs(u)

-1/T -fmax fmax 1/T u

Sampling Analysis: Freq. Domain

F(u)
 (u)
1
How about  (u − )
 T

-fmax fmax u … -1/T 0 1/T … u

F (u)   (u)

Fs(u)

-1/T -fmax fmax 1/T u

Sampling Analysis: Freq. Domain
+
n
F(u)   (u − )
n = − T


-fmax fmax u … -1/T 0 1/T … u

+
n
F (u )    (u − T )
n = −

Fs(u)

-1/T -fmax fmax 1/T u

 Sampling Theory
• How many samples are required to represent a given signal
without loss of information?

• What signals can be reconstructed without loss for a given

sampling rate?
 Sampling Analysis: Freq. Domain
+
n
F(u)   (u − )
n = − T


-fmax fmax u … -1/T 0 1/T … u

+
n
F (u )    (u − T )
n = −
How can we reconstruct the
original signal? Fs(u)

-1/T -fmax fmax 1/T u

Reconstruction in Freq. Domain
Fs(u) box(u)

-fmax fmax u -fmax fmax u

Fs (u )  box(u )

-fmax fmax u
Signal Reconstruction in Freq. Domain
Fs(u)
Fourier transform

x -fmax fmax u
… -2T -T 0 T 2T …
Fs (u )  box(u )
f(x)
Inverse Fourier F(u)
transform

x
-fmax fmax u

fs(x)
Signal Reconstruction in Spatial Domain
fs(x) sinc(x)

x
… -2T -T 0 T 2T …

f s ( x)  sin( x)

x
 Sampling Analysis
When does aliasing happen?

f(x) F(u)

x -fmax fmax u

fs(x) Fs(u)

x -fmax fmax u
… -T 0 T …
 Sampling Analysis
When does aliasing happen?
1
 2 f max
T

f(x) F(u)

x -fmax fmax u

fs(x) Fs(u)

x -fmax fmax u
… -T 0 T …
 Sampling Analysis
When does aliasing happen?
1
 2 f max
T

f(x) F(u)

x -fmax fmax u

fs(x) Fs(u)

x -fmax fmax u
… -T 0 T …
 Sampling Analysis
When does aliasing happen?
1
 2 f max
T

f(x) F(u)

x -fmax fmax u

fs(x) Fs(u)

x -fmax fmax u
… -T 0 T …
 Sampling Analysis
fs(x)
f(x)

Sampling

x Reconstruction x
… -2T -T 0 T 2T …

What sampling rate (T) is sufficient to reconstruct the

continuous version of the sampled signal?
 Sampling Analysis
fs(x)
f(x)

Sampling

x Reconstruction x
… -2T -T 0 T 2T …

What sampling rate (T) is sufficient to reconstruct the

continuous version of the sampled signal?
Sampling Rate ≥ 2 * max frequency in the signal
• this is known as the Nyquist Rate