DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

INDIAN INSTITUTE OF TECHNOLOGY (INDIAN SCHOOL OF MINES), DHANBAD

Winter Mid Semester Examination, Session 2022-23

Examination: VI BTech (CSE) Time: 2 Hours

Subject: Digital Image Processing (CsO304) Weightage: 32%, Max. Marks: 60

Instructions:
) Answer ALL questions.
i) Attempt all sub-parts of a same question at same place and in order.
ii) If you use any algorithm (covered in the class) to solve any problem, just mention it, do not elaborate.

Q. No. Question Mark

1.a For a unitary transform having a separable kernel, the forward transform can be expressed as 5
g T X f x7. Then the original image can be recovered involving unitary transform using
f Tx g xT.
Now, based on the above theorem, prove or disprove that the unitary transform work for the given image

by using the given unitary transform kernell

Consider an 8-level 8 x 8 input image as given below. Apply histogram equalization technique on the 5+5
input image and generate the frequency table for the output image.

3 446 5 2 0
5 7 1|5 4
5 1 5|2 6 4 1|
1 5 4 7 0
1 2 4 5 2 54
2 1 5 4 2 6 5|1
114 2 5 6 3 5
o2 5 0 4 55 0
Now, perform histogram specification using the target histogram (given below), where ni is the number
of pixels having i" graylevel and generate the frequency table for the output image.

0 1 2 34 5 6 T7
0 2 0 | 20 16 8

2.a Consider the given gray-level image of size 4x 4. Find the output
given image image when the
processed by 3 x 3 adaptive median filter. Note that the border pixels will remain unchanged. Showis6
all intermediate
steps of the executions of this filter clearly.
11 12 15 18
13 6543 44
16 48 14 34
13 24 22 27
Write a thinning algorithm with your own set of rules that transforms a square kernel (side length is 4 or 9
larger) to a straight line of size equal to 2 or larger. Consider that the input image is in
binary form.
 3. (a Consider the given gray-level image of size 4 x 4. Find the output image when the given image is 9

processed by 3 x 3 contraharmonic mean filter If(x, y )= 2(5.0)E.Sry 9(s,t)9*1

With Q = -1.5. Note that
the border
Z(6.1)5xy 9(s.t)0|
pixels remain unchanged. Show all intermediate steps of the
executions of this filter clearly.
1 2 2 5
2 135 143 4
4 158 164
3
4 6 1

b)Apply 1-D IDFT as per the formula given below to the given sequence of
x =
{20,-7+7j, -2,-7-7j}, where j v-1. Show all numbers:|6
intermediate steps clearly.
=

M-1
x)= M F(u)e j27Tux
M
u=0

4.(a Discuss the effect of using the

structuring element SE1 for boundary extraction of a binary image,
instead of the structuring element SE2 with a
proper visual example. The morphological equation to
extract the boundary of a binary
image A using a structuring element B is A -(A S B).

O110
SE1 SE2

(6) Explain qualitatively what effect the foiluwiig transtormations will have the
Transform of an image:
on Discrete Fourier 3+4

() convolution of the image with a Gaussian kernel of very tiny (nearly zero) standard deviation,
(i) point-wisemultiplication of the image with a complex exponential pattern
where a and b are known constants, and x, y are spatial coordinates.
exp(j2n(ax + by))

Justify or contrad the statement:

image enhancement in frequency
"Butterworth low pass filter is preferred to ideal low pass filter for | 2+1
domain". Under what condition the Butterworth low pass filter
becomes an ideal low pass filter?