Digital Signal and Image Processing

Contact Details:
Dr. Sushma Kadge
[email protected]
 Chapter 1

1.1 Introduction to digital signals and systems, Properties

and operations on digital signals.
1.2 Classification of signals, system, LTI system
1.3 Convolution in time domain (linear & circular),
Correlation.
Self-Learning Topic: Correlation (Circular)
 Signal
 A signal is a pattern of variation that carry information.

 Signals are represented mathematically as a function of one or

more independent variable.

 A picture is brightness as a function of two spatial variables, x and y.

 In this course signals involving a single independent variable,

generally refer to as a time, t are considered. Although it may not
represent time in specific application.

 A signal is a real-valued or scalar-valued function of an independent

variable t.

4
 Example of signals
 Electrical signals like voltages, current and EM field in
circuit
 Acoustic signals like audio or speech signals (analog or
digital)
 Video signals like intensity variation in an image
 Biological signal like sequence of bases in gene
 Noise which will be treated as unwanted signal

5
Classification of Signal
Continuous-time and discrete-time signal
Analog and digital signal (time and amplitude)
(1) Continuous-time signal：
(2) Discrete-time signal：Discrete variableContinuous amplitude
Time-domain discrete signals
(3) Analog Signal: Continuous variableContinuous amplitude
Speech, Television, Time-domain continuous signals
(4) Digital Signal：Discrete variablesDiscrete amplitude
Quantized discrete-time signals
 Signal Processing
Representation, transformation and manipulation
of signals and the information they contain.

Signal operation include:

(1) Transform, filter, inspection, spectrum analysis;
(2) Modulation and coding;
(3) Analog Signal Processing;
(4) Digital Signal Processing.

Computer, Semiconduct and Information Science

1960’s-1970’s
Processing of analog signal with digital methods

(1) Digitalized process for analog signals

xa(t) Sample Quantizer Coder x(n)

(2) Digital processing method

xa(t) Filter A/D DSP D/A Filter ya(t)

x(n) y(n)
Feature of Digital System
Advantages
(1) High accuracy: Floating point-8,16,32,64 bits
(2) High reliability: VLSI (analog: drift, calibration)
(3) Flexible: DSP, Software, FPGA, VHDL
(4) Easy to integrate
(5) Deal with high dimensional signals
(6) Low costs: reusable, reconfigurable
(7) Data logging
(8) Adaptive capability
Objective of Digital Signal Processing
Digital Signals

Manipulation Digital filter

Measurement Digital Signals

Spectrum analysis Frequency division
Disturbance attenuation
(1) Selective of A/D  Signal representation - Sampling

(2) Manipulation and transform  feature extraction and analysis

(3) Noise process  Digital filter

 Energy and Power
• Signals corrupted by noise.

• Energy and Power Spectrum useful to identify

signal from noise.

• Energy and Power spectrum can be calculated

by using Fourier transform.
 Power and Energy in a Physical System

 The instantaneous power

12

Pt
vvt
tit 
R

 The total energy

1

 
 

t t t
  

2 2 2 2
Ptdt vtitd
t vtd
t
t
1 t
1 t
1 R

 The average power

1t 1t 1
 
   
2 2 2
Ptdt vtdt
t
2t
1
t1 t
2t
1 R
t1

12
Power and Energy
 By definition, the total energy over the time interval t1  t  t2 in a
continuous-time signal x  t is:
 dt
t2
E xt
2

t1
x  t  denote the magnitude of the (possibly complex) number x  t 
 The time average power
1 t2
 
2
P xt dt
t2t
1
t
1

 By definition, the total energy over the time interval n2  n n1 in a

discrete-time signal x  n  is:
n2


Exn
2

nn1

 The time average power 1 n

 
2
2
P  xn
n
n
2 1 1
n
n1

13
14
• Discrete signal is function of discrete
independent variable.
• n is integer value of independent
variable.
• Discrete time signal
• Here independent variable is n along
with sampling time (T).

15
16
17
18
19
20
21
22
23
24
25
Sampling theorem
A continuous time signal can be
represented in its samples and can be
recovered back when sampling
frequency fs is greater than or equal to
the twice the highest frequency
component of message signal (fm).

26
Relationship between analog
and digital signal by sampling

27
28
Classification of Discrete Time
Signals

29
Deterministic and Random
Signal
• Signals specified by mathematical
equations are deterministic signals
Eg: Ramp , Unit Step

• Signals whose characteristics are

random in nature are random signals
Eg: Noise

30
Periodic and Non-periodic

• A discrete time signal is said to be

periodic if the signal remains the
same after left and right shifting
operation by (k.N) or T
• Where k is constant
• N is fundamental time period of the
signal
• Both k and N are integer values

31
32
33
34
35
36
• Even Signal: Discrete time signal
exhibits symmetry with respect to
n=0 it is even signal.
• Even signal satisfies this condition

37
38
• Odd Signal: Discrete time signal
exhibits antisymmetric with respect to
n=0 it is odd signal.
• Even signal satisfies this condition

39
40
• A discrete time signal is neither even
nor odd is expressed as :

41
• Calculation of Even and Odd part of
the signal:

42
43
44
• If energy of discrete time signal is
finite and non zero then the signal is
energy signal

45
• If power of discrete time signal is
finite and non zero then the signal is
power signal

46
47
48
• Causal Signal

• Non Causal if defined for n<=0 or for

both n>0 and n<=0
• If system is non causal then:

49
• Non causal discrete time signal only
defined for n<=0 is called anti causal
signal.

50
51
52
53
• Downsampling = signal x(n) here n
is replaced by Dn where D is an
integer..

• Upsampling = signal x(n) here n is

replaced by n/I where I is an
integer….

54
55
56
57
58
 Time Shifting of Discrete Time
Signals
• X(n) signal will be shifted by variable m
i.e. x(n)=x(n-m)
• m is positive integer signal is delayed by
m units of time.
• Delay results in shifting each sample of
x(n) to right.
• m is negative integer signal is advanced
by m units of time.
• Advance results in shifting each sample
of x(n) to left. 59
60
• Addition of two discrete signals is
performed on sample by sample
basis.

61
62
• Multiplication of two discrete signals
is performed on sample by sample
basis.

63
64
Discrete Time System
• Discrete time system is device or
algorithm which accepts input in
discrete signal form to produce output
or response in discrete time signal
form.

65
Discrete Time System

66
LTI (Linear Time Invariant)
System
• A discrete time system is linear if it
obeys principle of superposition and
time invariant if input output
relationship does not change with
time.

• When system satisfies property of

linearity and time invariance it is
called as LTI system.
67
Basic elements of Representation
are:
1) Adder
2) Constant Multiplier – Constant scaling
factor
3) Unit Delay Element – one sample unit
to delay
4) Unit Advance Element - one sample
unit to advance
68
69
70
71
72
• Static System – Output depends on
only current input samples but not on
past or future samples – no memory

• Dynamic System – Opposite of static

73
74
75
Time Invariant and Time
Variant System
• A system is said to be time invariant
if its input output characteristics do
not change with time.

76
Procedure to test for time
Invariance

77
Procedure to test for time
Invariance

78
Procedure to test for time
Invariance

79
80
81
82
Linear and Non Linear System
• Linear system satisfies superposition
principle.

• Response of system to weighted sum

= weighted sum of responses of the
system to each individual signals

83
Linear and Non Linear
System(Check for Linearity)

84
85
86
87
88
Causal and Non Causal Systems

89
Causal and Non Causal Systems

90
Causal and Non Causal Systems

91
Causal and Non Causal Systems

92
Stable and Unstable

93
Stable and Unstable

94
Stable and Unstable

95
FIR and IIR Systems
• FIR system (Finite duration Impulse
Response System) , the impulse
response consists of finite number of
samples.

• IIR system (Infinite duration Impulse

Response System) , the impulse
response consists of infinite number
of samples.
96
97
Discrete or Linear Convolution

98
Discrete or Linear Convolution
• In linear convolution the x1 and x2
sequences are non periodic , hence
convolution result x3 is also non
periodic.

• Linear convolution is said to be

aperiodic.

99
100
101
102
Number of Samples in output=N1+N2-1
N1 is number of samples in x(n)
N2 is number of samples in h(n)
Here 4+4-1= 7 samples for y(n)

103
104
Graphical Method

105
106
107
108
109
110
111
112
113
114
115
116
117
118
Circular Convolution

119
Circular Convolution

120
Circular Convolution

x(n) and h(n) --- Anticlockwise

H(n-1 ) --- Anticlockwise, shift one sample

H(n+1 ) --- Clockwise, shift one sample

H(-n) --- Clockwise

H(1-n ) --- Anticlockwise, shift one sample

121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
 Correlation
• Used when two signals are to be
compared.
• Measure of degree to which two
signals are similar.
• Cross Correlation – correlation of
two separate signals
• Auto Correlation – correlation of
signal with itself.

139
Cross Correlation

140
Auto Correlation

141
142
143
144
145
146
Matrix Method for Calculating
Auto and Cross Correlation.
• Steps for Matrix Method
Correlation
• Take signal x(n) as it is.
• Fold sequence y(n) to obtain y(-n).
• Obtain convolution of x(n) and y(-n).
• The result of convolution will be same
as correlation.

147