Digital Signal & Image Processing

Lecture-2
Dr Muhammad Arif
[email protected]
https://sites.google.com/site/mdotarif/teaching/dsip
 Overview
Digital Signal Processing System
Analog to Digital Conversion
NyquistShannon Sampling Theorem
Aliasing
Sampling Effect in Time Domain
Sampling Effect in Frequency Domain
Anti Aliasing Filter
Under-sampling
Sampling of Band Limited Signals
Over-sampling
Digital to Analog Conversion
 Analog vs. Digital Signal Processing

Analog input Signal x(t) Analog output Signal y(t)

Analog
Signal Processor

Analog Signal Processing

Analog input Analog output

Signal x(t) Signal y(t)
A/D Digital D/A
converter Signal Processor converter

Digital Signal Processing

3
Typical Digital Signal Processing System

It consists of
an analog filter called (anti-imaging) filter,
an analog-to-digital conversion (ADC) unit,
a digital signal (DS) processor,
a digital-to-analog conversion (DAC) unit,
and an analog filter called reconstruction (anti-image) filter.

4
Typical Digital Signal Processing System

5
A/D & D/A Conversion

6
 Analog to Digital (A/D) Conversion

Most signals of practical interest are analog in nature

Examples: Voice, Video, RADAR signals, Transducer/Sensor
output, Biological signals etc

So in order to utilize those benefits, we need to convert our

analog signals into digital

This process is called A/D conversion

7
 Analog to Digital Conversion
A/D conversion can be viewed as a three step process

8
 Analog to Digital Conversion
A/D conversion can be viewed as a three step process

9
 Analog to Digital Conversion
Sample & Hold (Sampler)

Analog signal is continuous in time and continuous in

amplitude.

It means that it carries infinite information of time and infinite

information of amplitude.

Analog (continuous-time) signal has some value defined at

every time instant, so it has infinite number of sample points.

10
 Analog to Digital Conversion
Sample & Hold (Sampler)

It is impossible to digitize an infinite number of points.

The infinite points cannot be processed by the digital signal

(DS) processor or computer, since they require an infinite
amount of memory and infinite amount of processing power
for computations.

Sampling is the process to reduce the time information or

sample points.

11
 Analog to Digital Conversion
Sample & Hold (Sampler)

The first essential step in analog-to-digital (A/D) conversion is

to sample an analog signal.

This step is performed by a sample and hold circuit, which

samples at regular intervals called sampling intervals.

Sampling can take samples at a fixed time interval.

The length of the sampling interval is the same as the

sampling period, and the reciprocal of the sampling period is
the sampling frequency fs.
12
 Analog to Digital Conversion
Sample & Hold (Sampler)

After a brief acquisition time, during which a sample is

acquired, the sample and hold circuit holds the sample steady
for the remainder of the sampling interval.
The hold time is needed to allow time for an A/D converter to
generate a digital code that best corresponds to the analog
sample.
If x(t) is the input to the sampler, the output is x(nT), where T
is called the sampling interval or sampling period.
After the sampling, the signal is called discrete time
continuous signal which is discrete in time and continuous in
amplitude.
13
Analog to Digital Conversion
Sample & Hold (Sampler)

14
 Analog to Digital Conversion
Sample & Hold (Sampler)
Figure below shows an analog (continuous-time) signal (solid
line) defined at every point over the time axis (horizontal line)
and amplitude axis (vertical line).
Hence, the analog signal contains an infinite number of points.

15
 Analog to Digital Conversion
Sample & Hold (Sampler)
Each sample maintains its voltage level during the sampling
interval to give the ADC enough time to convert it.
This process is called sample and hold.

16
NyquistShannon Sampling Theorem

The sampling theorem guarantees that an analogue signal can be

perfectly recovered as long as the sampling rate is at least twice
as large as the highest-frequency component of the analogue
signal to be sampled.

17
NyquistShannon Sampling Theorem

18
NyquistShannon Sampling Theorem

Examples

19
 NyquistShannon Sampling Theorem
Example: For the following analog signal, find the Nyquist sampling
rate, also determine the digital signal frequency and the digital
signal

20
NyquistShannon Sampling Theorem

Example: Find the sampling frequency of the following signal.

So sampling frequency should be

21
NyquistShannon Sampling Theorem

Exercise

Determine the Nyquist sampling rate of a signal

x(t) = 3sin(5000t + 17o)

22
Aliasing

23
 Aliasing
How many hertz can the human eye see?

Most don't notice unless it is under 50 or 60 Hz.

Generally, people notice when the frame-rate is less than the

refresh rate of the display.

Depending on the type of CRT, you couldn't see flicker at 30

Hz or you could still see it at 120 Hz.

24
 Aliasing
When the minimum sampling rate is not respected, distortion
called aliasing occurs.

Aliasing causes high frequency signals to appear as lower

frequency signals.

To be sure aliasing will not occur, sampling is always preceded

by low pass filtering.

The low pass filter, called the anti-aliasing filter, removes all
frequencies above half the selected sampling rate.

25
 Aliasing
Figure illustrates sampling a 40 Hz sinusoid
The sampling interval between sample points is T = 0.01 second,
and the sampling rate is thus fs = 100 Hz.
The sampling theorem condition is satisfied

26
 Aliasing
Figure illustrates sampling a 90 Hz sinusoid
The sampling interval between sample points is T = 0.01 second,
and the sampling rate is thus fs = 100 Hz.
The sampling theorem condition is not satisfied

27
Aliasing

28
 Sampling Effect in Time Domain

Example of Aliasing in the time domain

of various sinusoidal signals ranging
from 10 kHz to 80 kHz with a sampling
frequency Fs = 40 kHz.

29
 Time & Frequency Domains
There are two complementary signal descriptions.
Signals seen as projected onto time or frequency domains.

30
Time & Frequency Domains

31
Signal & Spectrum

32
Frequency Range of Analog & Digital Signals

For analog signals, the frequency range is from - Hz to Hz

For digital signals, the frequency range is from 0 Hz to Fs/2 Hz

33
Sampling Effect in Frequency Domain

Sampling causes images of a signals spectrum to appear at

every multiple of the sampling frequency fs.

For a signal with frequency f, the sampled spectrum has

frequency components at kfs f

34
Sampling Effect in Frequency Domain

35
Sampling Effect in Frequency Domain

36
 Anti Aliasing Filter
A signal with no frequency component above a certain
maximum frequency is known as a band-limited signal.

In our case we want to have a signal band-limited to Fs.

Some times higher frequency components (both harmonics

and noise) are added to the analog signal (practical signals are
not band-limited).

In order to keep analog signal band-limited, we need a filter,

usually a low pass that stops all frequencies above Fs.

This is called an Anti-Aliasing filter. 37

 Anti Aliasing Filter
Anti-aliasing filters are analog filters.

They process the signal before it is sampled.

In most cases, they are also low-pass filters unless band-pass

sampling techniques are used.

38
 Under Sampling
If the sampling rate is lower than the required Nyquist rate, that
is fS < 2W, it is called under sampling.

In under sampling images of high frequency signals erroneously

appear in the baseband (or Nyquist range) due to aliasing.

39
 Sampling of Band Limited Signals

Signals whose frequencies are restricted to a narrow band of

high frequencies can be sampled at a rate similar to twice the
Bandwidth (BW) instead of twice the maximum frequency.

Fs BW

40
 Sampling of Band Limited Signals
While this under-sampling is normally avoided, it can be
exploited.
For example, in the case of band limited signals all of the
important signal characteristics can be deduced from the copy
of the spectrum that appears in the baseband through
sampling.
Depending on the relationship between the signal frequencies
and the sampling rate, spectral inversion may cause the shape
of the spectrum in the baseband to be inverted from the true
spectrum of the signal.

41
 Sampling of Band Limited Signals

Figure: Signal recovered

From Nyquist range are
Base band versions of the
Original signal. Sampling rate is
Important to make sure no aliasing
and spectral inversion occurs.

(a) Fs = 80 kHz, signal spectrum

is Inverted in the baseband.

(b) Fs = 100 kHz, the lowest

Frequencies In the signal alias
to the highest frequencies.

(c) Fs = 120 kHz, No spectral

42
Inversion occurs.
 Over Sampling
Oversampling is defined as sampling above the minimum
Nyquist rate, that is, fS > 2fmax.

Oversampling is useful because it creates space in the

spectrum that can reduce the demands on the analog anti-
aliasing filter.

43
 Over Sampling
In the example below, 2x oversampling means that a low order analog filter is
adequate to keep important signal information intact after sampling.
After sampling, higher order digital filter can be used to extract the information.

44
 Over Sampling
The ideal filter has a flat pass-band and the cut-off is very sharp,
since the cut-off frequency of this filter is half of that of the
sampling frequency, the resulting replicated spectrum of the
sampled signal do not overlap each other. Thus no aliasing
occurs.
Practical low-pass filters cannot achieve the ideal
characteristics.
Firstly, this would mean that we have to sample the filtered
signals at a rate that is higher than the Nyquist rate to
compensate for the transition band of the filter

45
 Spectra of Sampled signals

Figure: Signal s Spectra

(i) Over sampled
(ii) Nyquest Rate
(iii) Under Sampled

46
Sampling Low Pass Signals

47
 Exercise
Exercise-1: If the 20 kHz signal is under-sampled at 30 kHz, find the aliased
frequency of the signal.

Exercise-2: A voice signal is sampled at 8000 samples per second.

i. What is the time between samples?
ii. What is the maximum frequency that will be recovered from the signal?

Exercise-3: An analog Electromyogram (EMG) signal contains useful

frequencies up to 3000 Hz.
i. Determine the minimum required sampling rate to avoid aliasing.
ii. Suppose that we sample this signal at a rate of 6500 samples/s. what is
the highest frequency that can be represented uniquely at this sampling
rate?

48
 Exercise
Exercise-4: Humans can hear sounds at frequencies between 0 and 20 kHz.
What minimum sampling rate should be chosen to permit perfect recovery
from samples?

Exercise-5: An ECG signal is sampled at 250 samples per second.

i. What is the time between samples?
ii. What is the maximum frequency that will be recovered from the signal?

Exercise-6: An ultrasound signal ranging in frequency from 900 kHz to 900.5

kHz is under-sampled at 200 kHz. If a 200 Hz target appears in the baseband,
what is the actual frequency of the target?

49
 Analog to Digital Conversion
Quantizer

After the sampling, the discrete time continuous signal still

carry infinite information (can take any value) in terms of
amplitude.

Quantization is the process to reduce infinite information of

the amplitude.

Quantizer do the conversion of discrete time continuous

valued signal into a discrete-time discrete-value signal.

The value of each signal sample is represented by a value

selected from a finite set of possible values.
50
 Analog to Digital Conversion
Quantizer

The A/D converter chooses a quantization level for each

analog sample.

Number of levels of quantizer is equal to L = 2N

An N-bit converter chooses among 2N possible quantization

levels.

So 3 bit converter has 8 quantization levels, and 4 bit

converter has 8 quantization levels.
51
 Analog to Digital Conversion
The quantization step size or resolution is calculated as:
= Q = R/2N
where
R is the full scale range of the analog signal (i.e. Ymax - Ymin)
N is the number of bits used by the converter

Resolution of a quantizer is the distance between two

successive quantization levels
More quantization levels, a better resolution!
What's the downside of more quantization levels?

The strength of the signal compared to that of the quantization

errors is measured by dynamic range and signal-to-noise ratio.
52
Analog to Digital Conversion
4-bit Quantizer

53
Analog to Digital Conversion
4-bit Quantizer

54
 Quantization Error
The error caused by representing a continuous-valued signal
(infinite set) by a finite set of discrete-valued levels.

The larger the number of quantization levels, the smaller the

quantization errors.

The quantization error is calculated as the difference between

the quantized level and the true sample level.

Most quantization errors are limited in size to half a

quantization step Q or .
55
 Quantization Error
Suppose a quantizer operation given by Q(.) is performed on
continuous-valued samples x[n] is given by Q(x[n]), then the
quantization error is given by

56
 Analog to Digital Conversion
Lets consider the signal which is to be quantized.

In the figure, we can see that there is a difference between the

original signal (Blue Line) and the quantized signal (Red Lines).
This is the error produced while quantization
57
 Analog to Digital Conversion
Quantization error can be reduced, however, if the number of
quantization levels is increased as illustrated in the figure

58
Analog to Digital Conversion
Quantization of unipolar data (maximum error = full step)

59
Analog to Digital Conversion
Quantization of unipolar data (maximum error = half step)

60
 Analog to Digital Conversion
Example: Analog pressures are recorded using a pressure transducer as
voltages between 0 and 3 V. The signal must be quantized using a 3-bit
digital code. Indicate how the analog voltages will be covered to digital
values.

The quantization step size is

Q = 3 V/23 = 0.375 V

The half of quantization step is

0.1875 V

61
Analog to Digital Conversion
Quantization of bipolar data (maximum error = half step)

62
Three-bit A/D Conversion

63
 Dynamic Range
Quantization errors can be determined by the quantization
step.
Quantization errors can be reduced by increasing the number
of bits used to represent each sample.
Unfortunately these errors can not be entirely eliminated and
their combined effect is called quantization noise.

The dynamic range of the quantizer is the number of levels it

can distinguish in noise.
It is a function of the range of signal values and the range of
error values, and is expressed in decibels, dB.

= 202

64
 Signal-to-Quantization-Noise Ratio

Provides the ratio of the signal power to the quantization

noise (or quantization error)

Mathematically,

where
Px= Power of the signal x (before quantization)
Pq= Power of the error signal xq
65
Analog to Digital Conversion

66
2 bit Flash ADC

67
Digital-to-Analog (D/A) Conversion

Block Diagram of D/A Conversion

68
 Digital-to-Analog (D/A) Conversion

Once digital signal processing is complete, digital-to-analog

(D/A) conversion must occur.
This process begins by converting each digital code into an
analog voltage that is proportional in size to the number
represented by the code.
This voltage is held steady through zero order hold until the next
code is available, one sampling interval later.
This creates a staircase-like signal that contains frequencies
above W Hz.
These signals are removed with a smoothing analog low pass
filter, the last step in D/A conversion.
69
 Digital-to-Analog (D/A) Conversion

In the frequency domain, the high frequency elements

present in the zero order hold signal appear as images, copies
of the original signal spectrum situated around integer
multiples of the sampling frequency.

The smoothing analog filter removes these images and so is

given the name of Anti-Imaging Filter.

Only the frequencies in the baseband, between 0 and fS/2 Hz,

remain.

70
Digital-to-Analog (D/A) Conversion
Three bit D/A Conversion

71
Comparing Signals in the A/D & D/A Chain

72
Comparing Signals in the A/D & D/A Chain

73
Summary

74
Summary

75