Digital Image Processing Assignment-Week 2: NPTEL Online Certification Courses Indian Institute of Technology Kharagpur
Digital Image Processing Assignment-Week 2: NPTEL Online Certification Courses Indian Institute of Technology Kharagpur
QUESTION 1:
A signal has a bandwidth of 4 KHz. If every sample is digitized using 16 bits and the digital speech
is to be transmitted over a communication channel, what is the minimum bandwidth requirement of
the channel?
a) 16 Kbps
b) 32 Kbps
c) 64 Kbps
d) 128 Kbps
Detailed Solution:
First we need to convert continuous signal into discrete signal by the process of sampling and then
digitize using digital bits.
QUESTION 2:
100 150 70 50 50 25
Consider the following two images f1 = ( 50 55 175) and f2 = (45 55 50).
200 50 150 50 50 75
The addition and subtraction of images are given by f1 + f2 and f1 − f2 . Assume both the images
are of the 8-bit integer type.
51 153 132 −49 −147 −118
a. f1 + f2 = ( 50 70 230) and f1 − f2 = (−40 −40 20 )
255 100 225 0 0 75
51 153 132 0 0 0
b. f1 + f2 = ( 50 70 255) and f1 − f2 = (0 0 20)
255 100 225 0 0 75
51 153 132 0 0 0
d. f1 + f2 = ( 50 70 230) and f1 − f2 = (0 0 20)
255 100 225 0 0 75
Detailed Solution:
QUESTION 3:
In the following 3x3 image I(x, y), which of the option is true?
Correct Answer: b.
Detailed Solution:
There exist a 4-connectivity between I(1, 1) and I(0, 1) and therefore these two are not m-
connected. But from definition of 8-connectivity, I(1, 1) and I(0, 1) are 8-connected.
QUESTION 4:
Find the value of logical operation OR for the binary images A and B (A OR B). Assume 1 to be
foreground and 0 to be background pixels.
Image-A Image-B
Correct Answer: C.
Detailed Solution:
OR between and is
QUESTION 5:
Compute the Euclidean Distance (D1), City-block Distance (D2) and Chessboard distance (D3) for
points p and q, where p and q be (1, 2, 3) and (1, 5, 7) respectively. Give answer in the form (D 1,
D2, D3).
a) (5, 3, 4)
b) (5, 4, 3)
c) (7, 5, 4)
d) (5, 7, 4)
Detailed Solution:
QUESTION 6:
Consider an 3-D point [5 5 5]T. Perform a scaling operation (S) in both x-axis and y-axis by 5 units.
Detailed Solution:
matrix in homogenous
system
X 5 0 0 0 5
Y 0 5 0 0 5
Z 0 0 1 0 5
1 0 0 0 1 1 .
On solving we get,
X 25 , Y 25 and Z 5 .
QUESTION 7:
The output of an image sensor takes values between 0.0 and 20.0. If it is quantized by a uniform
quantizer with 256 levels, what will be the transition ( t k ) and reconstruction level ( rk )?
10(k 1) 5
a) t k , k 1,2,3,....,257; rk t k , k 1,2,3,....,256
256 256
10(k 1) 10
b) t k , k 1,2,3,....,257; rk t k , k 1,2,3,....,256
256 256
10(k ) 5
c) t k , k 1,2,3,....,257; rk t k , k 1,2,3,....,256
256 256
20(k 1) 10
d) t k , k 1,2,3,....,257; rk t k , k 1,2,3,....,256
256 256
20(k 1) 10
Correct Answer: d. t k , k 1,2,3,....,257; rk t k , k 1,2,3,....,256 .
256 256
Detailed Solution:
20
Using the uniform quantizer design, q , then the transition and reconstruction levels can be
256
defined as
20(k 1) q 10
. t k t k 1 q , k 1,2,3,....,257; rk t k t k , k 1,2,3,....,256
256 2 256
Hence option d is correct.
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur
QUESTION 8:
Consider an image point [1,1]T in a continuous image. Rotation of the image point around the
origin by 45 o in anticlockwise direction around origin is given by :
a) [0, √2]T
b) [0, 2]T
c) [1, 2√2]T
d) [1, √2]T
Detailed Solution:
QUESTION 9:
5 5 5 5 5
5 5 5 5 5
Find the 2D convolution of the given matrices using zero padding. Input = 5 5 5 5 5
10 10 10 10 10
10 10 10 10 10
1 2 1
and kernel = 0 0 0 . (Discard padded positions so that final answer will be a 5x5 matrix)
1 2 1
15 20 20 20 15
0 0 0 0 0
a) 15 20 20 20 15
15 20 20 20 15
30 40 40 40 30
15 20 20 20 15
0 0 0 0 0
b) 15 20 20 20 15
15 20 20 20 15
30 40 40 40 30
15 20 20 20 15
0 0 0 0 0
c) 15 20 20 20 15
15 20 20 20 15
30 40 40 40 30
15 20 20 20 15
0 0 0 0 0
d) 15 20 20 20 15
15 20 20 20 15
30 40 40 40 30
15 20 20 20 15
0 0 0 0 0
Correct Answer: a. 15 20 20 20 15
15 20 20 20 15
30 40 40 40 30
Detailed Solution:
For convolution, we have to do either time reversal of input image or the kernel. Let’s perform time
reversal of kernel and move the kernel on the input image.
5 5 5 5 5
5 5 5 5 5
1 2 1
Input image, I = 5 5 5 5 5 and time reversed kernel= 0 0 0
10 10 10 10 10 1 2 1
10 10 10 10 10
I(0,0) = 0x1 + 0x2 + 0x1 + 0x0 + 5x0 + 5x0 + 0x(-1) + 5x(-2) + 5x(-1) = -15
I(0,1) = 0x1 + 0x2 + 0x1 + 5x0 + 5x0 + 5x0 + 5x(-1) + 5x(-2) + 5x(-1) = -20
In this we move the time reversed kernel over the image I and finally obtain the convolved image
as
15 20 20 20 15
0 0 0 0 0
15 20 20 20 15
15 20 20 20 15
30 40 40 40 30
QUESTION 10:
For a B-bit uniform optimal qauntizer which of the following relation is true for signal to noise
ratio (SNR)?
a) SNR 10 log 10 2 2 B
b) SNR 10 log 10 2 B
c) SNR 10 log 10 2 B
d) SNR 10 log 10 2 2 B
Correct Answer: d.
Detailed Solution:
A2
The variance, of a uniform random variable, whose range is A, is given by
2
u . Also for a
12
uniform quantizer with B bits, where every level has to be represented by B bits, we will have
A q2 u2
q B . The MSE error due to quantization is given by . SNR is nothing but . On
2 12
simplification we have SNR 10 log 10 2 2 B .