Nptel Online Certification Course: Computer Vision

Assignment 1: Fundamentals of Image processing

Number of questions: 10 Total marks: 10 X 2 = 20

QUESTION 1: Type:MCQ
Which transform is most commonly used in frequency domain image processing?

a) Laplace Transform

b) Fourier Transform

c) Z-Transform

d) Hadamard Transform

Correct Answer: b)

QUESTION 2: Type:MCQ
Note: Note that the probability of a normal variable P(z > 2.0) = 0.0228
X is a normally distributed variable with mean µ = 30 and standard deviation σ = 4.
Compute P(26 < X < 38).

a) 0.0238

b) 0.9380

c) 0.8185

d) 0.9772

Correct Answer: c)
Detailed Solution: For X = 26, the z value is z1 = (X − µ)/σ = (26 − 30)/4 = −1.0. For
X = 38, the z value is z1 = (X − µ)/σ = (38 − 30)/4 = 2.0. Hence
P(X < 26)=P(z < −1.0)=0.1587 and P(X > 38)=P(z > 2.0)=0.0228, which means
P(26 < X < 38)=1 − P(X < 26) − P(X > 38) = 1 − 0.1587 − 0.0228 = 0.8185.

1
QUESTION 3: Type:MCQ
Consider the following 5 × 5 image array to answer the following question.
 
5 6 4 6 4

2 2 0 3 4

1 2 4 0 5

5 1 3∗ 2 1
 

5 2 6 4 6
What is the value after histogram equalization at location (3,2) shown with asterisk mark in
the image array.

a) 2

b) 3

c) 4

d) 5

Correct Answer: b)
Detailed Solution:
Computation of Histogram equalization is shown in the following table.

pixel frequency cumulative sum Normalized sum New pixel

0 2 2 0.48 0
1 3 5 1.2 1
2 5 10 2.4 2
3 2 12 2.88 3
4 5 17 4.08 4
5 4 21 5.04 5
6 4 25 6.0 6

According to the above table pixel value 3 remains 3. Hence, 3 is the answer.

2
QUESTION 4: Type:MCQ
A continuous time signal is given by x(t) = e u(t), its fourier transform X( jω) is given by
−2t

a) 1/(1 + jω)

b) 1/(2 + jω)

c) 1/(3 + jω)

d) 1/(1 + 2 jω)

Correct Answer: b)
Detailed RSolution:
∞ R∞
X( jω) = −∞ e−2t u(t)e− jwt dt = 0 e−2+ jω dt = −1/(2 + jω)(e−∞ − e0 ) = 1/(2 + jω).

QUESTION 5: Type:MCQ
Consider two images each of size 32x32, namely I1 and I2 . Image I1 has 612 and 412 pixels
for foreground and background, respectively. Image I2 has 412 and 612 pixels for foreground
and background, respectively. One pixel is selected at random from one of the two images
and it is found to be a background pixel. Find the probability that it was drawn from image
I1 ?

a) 0.4

b) 0.04

c) 0.6

d) 0.24

Correct Answer: a)
Detailed Solution:
Let E1 and E2 are the events of choosing images I1 and I2 respectively. Let A be the event of
selecting the background pixel. Now, use Bayes theorem to evaluate p(E1 |A), where p(E1 ),
p(E2 ), p(A|E1 ) and p(A|E2 ) are 0.5, 0.5, 0.4 and 0.6, respectively. The probability that
background pixel is drawn from image I1 is p(E1 |A) = p(A/E1 ) × p(E1 )/ (p(A/E1 ) × p(E1 ) +
p(A/E2 ) × p(E2 )).

3
QUESTION 6: Type:MSQ
Which of the following statements about image processing fundamentals are correct?

a) Grayscale conversion reduces an image to a single channel by averaging RGB values.

b) Histogram equalization increases contrast by redistributing pixel intensities non-linearly.

c) Frequency-domain filtering (e.g., Fourier Transform) modifies an image by altering its

frequency components.

d) Morphological operations (e.g., erosion/dilation) are only applicable to binary images.

Correct Answer: a), c)

QUESTION 7: Type: MCQ

Consider the following 5 × 5 image array and 3 × 3 sobel operator to answer the following
question.
3 4 5 6 7
2 1 0 3 1 1 2 1
∗
5 9 6 0 1 0 0 0
1 4 3 2 1 -1 -2 -1
7 2 1 8 9
What is the output of 3 × 3 sobel operator at location (2, 2) shown with asterisk mark in the
image array?

a) -8

b) -19

c) -13

d) 17

Answer: a)
Detailed Solution:  
1 0 3
3x3 matrix from the given location 9 6 0
 
4 3 2
 
 
 1 2 1 
Sobel Kernel  0 0 0  Do elementary multiplication of these matrices and take the sum.
 
−1 −2 −1
 

4
QUESTION 8: Type:MCQ
Consider the following 3-bit grey scale image

0 1 0 3
4 0 6 7
0 3 2 5
4 1 0 6

What is the threshold according to Otsu method?

a) 2

b) 3

c) 4

d) 7

Correct Answer: a)
Detailed Solution: The grey level 2, maximizes the between class variance.

QUESTION 9: Type: Comprehensive

Consider a 10-bit grey scale image with dimension 32 × 32. What will be the range of
values in the X-axis of its histogram?

a) 0 to 1024

b) 1 to 1024

c) 0 to 1023

d) 1 to 1023

Correct Answer: c)
Detailed Solution:
For a 10-bit image, there are 210 intensities. The intensities start from 0. So, the range is 0 to
1023.

5
QUESTION 10: Type: MCQ
Consider the following 3-bit grey scale image

0 1 4 3
2 4 3 0
0 1 2 7
4 1 5 6

What is the value when vertical Prewitt operator and horizontal Prewitt operator are applied
on the orange colored pixel?

a) 1 and −4

b) 4 and 2

c) 4 and 1

d) −4 and 1

Correct Answer: c)
Detailed Solution:
Vertical Prewitt Operator (Horizontal Edge Detection)
 
−1 0 1
G x = −1 0 1
 
−1 0 1
 

Horizontal Prewitt Operator (Vertical Edge Detection)

 
−1 −1 −1
Gy =  0 0 0 
 
1 1 1
 