General Knowledge

1. Describe within three lines the reason of occurrence of color blindness.

2. Write a python script that mimics the color blindness for 100 frames after reading them
from default webcam.

3. While applying an averaging filter or weighted averaging filter to the image, we divide
the averaging window by sum of all the elements of that window. Why we perform such
division? What will happen if we don’t apply such division? Answer within six lines.

4. Which histogram represents the tire image? And why? Your answer should be within
three lines.

A B C
5. For the images of sizes 8x8 shown in figure below.
a. What will be the difference of histograms of both images?
b. Sketch the histogram of both images.
Note: The possible value of each pixel is either 0 or 255 in given figures.

6. Describe the difference between geometric transformation and intensity transformation

within three lines.
7. A common measure of transmission for digital data is the baud rate, defined as the
number of bits transmitted per second. Generally, transmission is accomplished in
packets consisting of a start bit, a byte (8 bits) of information, and a stop bit. Using these
facts, calculate that how many minutes would it take to transmit a 1024x1024 image with
16 intensity levels using a 56K baud modem?
8. A common measure of transmission for digital data is the baud rate, defined as the
number of bits transmitted per second. Generally, transmission is accomplished in
packets consisting of a start bit, a byte (8 bits) of information, and a stop bit. Using these
facts, calculate that how many seconds would it take to transmit a 1024x1024 image with
16 intensity levels at 3000K baud, a representative medium speed of a phone DSL
(Digital Subscriber Line) connection?
9. A common measure of transmission for digital data is the baud rate, defined as the
number of bits transmitted per second. Generally, transmission is accomplished in
packets consisting of a start bit, a byte (8 bits) of information, and a stop bit. Using these
facts, calculate that how many minutes would it take to transmit a 1024x1024 image with
256 intensity levels using a 56K baud modem?
10. A common measure of transmission for digital data is the baud rate, defined as the
number of bits transmitted per second. Generally, transmission is accomplished in
packets consisting of a start bit, a byte (8 bits) of information, and a stop bit. Using these
facts, calculate that how many seconds would it take to transmit a 1024x1024 image with
256 intensity levels at 3000K baud, a representative medium speed of a phone DSL
(Digital Subscriber Line) connection?
11. A common measure of transmission for digital data is the baud rate, defined as the
number of bits transmitted per second. Generally, transmission is accomplished in
packets consisting of a start bit, a byte (8 bits) of information, and a stop bit. Using these
facts, calculate that how many minutes would it take to transmit a 1024x1024 binary
image using a 56K baud modem?
12. A common measure of transmission for digital data is the baud rate, defined as the
number of bits transmitted per second. Generally, transmission is accomplished in
packets consisting of a start bit, a byte (8 bits) of information, and a stop bit. Using these
facts, calculate that how many seconds would it take to transmit a 1024x1024 binary
image at 3000K baud, a representative medium speed of a phone DSL (Digital Subscriber
Line) connection?
13. If we have two histograms h1 and h2 that are same. Do they necessarily represent the
same image? Provide reason of your answer.
14. Describe the difference between digital representation of color and grayscale image.
What information do the red, green and blue channels provide? What will happen if we
set all the values of green channel to zero?
15. Which color scheme is used in printing press and which one is similar to human visual
system?
16. What type of operator is used to detect an isolated point? Why it provides maximum
response on isolated point.
17. Suppose that someone took a picture with very poor lightening condition (image is dark).
Name three algorithms that can be used to correct the illumination of the image.
18. When you enter a dark theater on a bright day, it takes an appreciable interval of time
before you can see well enough to find an empty seat. Which of the visual processes
explained in the first week of this course is at play in this situation? Briefly explain the
reason as well.
19. Alternating current certainly is part of the electromagnetic spectrum. Commercial
alternating current in the United States has a frequency of 60 Hz. What is the wavelength
in kilometers of this component of the spectrum?
20. Image subtraction is used often in industrial applications for detecting missing
components in product assembly. The approach is to store a “golden” image that
corresponds to a correct assembly; this image is then subtracted from incoming images of
the same product. Ideally, the differences would be zero if the new products are
assembled correctly. Difference images for products with missing components would be
nonzero in the area where they differ from the golden image. What conditions do you
think have to be met in practice for this method to work?
21. What is the non-maximal suppression technique used in Canny Algorithm. How does it
work?
22. What role does thresholding plays in edge detection?

Image Denoising Section:

23. Suppose that you have a row that represent a 1D ramp edge with the following intensities
20 20 20 30 40 50 6070 80 80 80 80
Apply 1x3 mean and 1x3 weighted mean filter. Which filter perform well in preserving
the ramp edge and why?

24. Suppose that you have a row that represent a 1D ramp edge with the following intensities
20 20 20 30 40 50 6070 80 80 80 80
Apply 1x3 mean and 1x3 median filter. Which filter perform well in preserving the ramp
edge and why?

25. Suppose that you have a row that represent a 1D ramp edge with the following intensities
20 20 20 30 40 50 6070 80 80 80 80
Apply 1x3 mean and 1x5 mean filter. Which filter perform well in preserving the ramp
edge and why?

26. Suppose that you have a row that represent a 1D ramp edge with the following intensities
20 20 20 30 40 50 6070 80 80 80 80
Apply 1x3 median and 1x5 median filter. Which filter perform well in preserving the
ramp edge and why?
27. Suppose that you have a row that represent a 1D roof edge with the following intensities
20 20 20 30 40 50 6070 80 70 60 50 40 30 20 20 20
Apply 1x3 mean and 1x5 mean filter. Which filter perform well in preserving this edge
and why?

28. Suppose that you have a row that represent a 1D roof edge with the following intensities
20 20 20 30 40 50 6070 80 70 60 50 40 30 20 20 20
Apply 1x3 median and 1x5 median filter. Which filter perform well in preserving this
edge and why?

29. Suppose that you have a row that represent a 1D step edge with the following intensities
20 20 20 20 80 80 80 80 80
Apply 1x3 mean and 1x3 weighted mean filter. Which filter perform well in preserving
this edge and why?

30. Suppose that you have a row that represent a 1D step edge with the following intensities
20 20 20 20 80 80 80 80 80
Apply 1x3 mean and 1x3 median filter. Which filter perform well in preserving this edge
and why?

31. Suppose that you have a row that represent a 1D step edge with the following intensities
20 20 20 20 80 80 80 80 80
Apply 1x3 mean and 1x5 mean filter. Which filter perform well in preserving this edge
and why?

32. Suppose that you have a row that represent a 1D step edge with the following intensities
20 20 20 20 80 80 80 80 80
Apply 1x3 median and 1x5 median filter. Which filter perform well in preserving this
edge and why?

33. The difference of grayscale and color image in term of image representation is that the
color image is represented using 3D array whereas grayscale is represented using 2D
array. Imagine if you have a color image represented with a 3D Array. Suggest some
changes to that 3D array in such a way that the image looks grayscale image while still
represented with 3D array

34. The white bars in the image above are 7 pixels wide and 210 pixels high. The separation
between bars is 17 pixels. What would this image look like after application of
a. A 3x3 arithmetic mean filter?
b. A 7x7 arithmetic mean filter?
c. A 9x9 arithmetic mean filter?

35. The white bars in the image above are 7 pixels wide and 210 pixels high. The separation
between bars is 17 pixels. What would this image look like after application of
a. A 3x3 geometric mean filter?
b. A 7x7 geometric mean filter?
c. A 9x9 geometric mean filter?

36. The white bars in the image above are 7 pixels wide and 210 pixels high. The separation
between bars is 17 pixels. What would this image look like after application of
a. A 3x3 median filter?
b. A 7x7 median filter?
c. A 9x9 median filter?
37. The white bars in the image above are 7 pixels wide and 210 pixels high. The separation
between bars is 17 pixels. What would this image look like after application of
a. A 3x3 max filter?
b. A 7x7 max filter?
c. A 9x9 max filter?

38. The white bars in the image above are 7 pixels wide and 210 pixels high. The separation
between bars is 17 pixels. What would this image look like after application of
a. A 3x3 mid-point filter?
b. A 7x7 mid-point filter?
c. A 9x9 mid-point filter?
39. The subimage on the left is the result of using an arithmetic mean filter of size 3x3 the
other subimage is the result of using a geometric mean filter of the same size.
a. Explain why the subimage obtained with geometric mean filtering is less blurred.
(Hint: Start your analysis by examining a 1-D step transition in intensity.)
b. Explain why the black components in the right image are thicker.

40. If we convolve the following image with the simple smoothing filter of size (3x3), what
would the output Image G look like? Consider padding with white pixels?
41. The figure on the left consists of two halves: the left half includes black pixels with a few
white pixels, and the right half includes white pixels with a few black pixels. If the goal is
to remove the noisy pixels from both halves such that we need to remove the three white
pixels from the left half and the four white pixels from the right half as shown in the right
figure, describe a mechanism to achieve this goal? Answer within five lines.

42. The equation of alpha trimmed mean filter is given below

1
fˆ ( x, y )   g ( s, t ) r
mn  d ( s ,t )S xy

Suppose that we have the following 5x5 patch corrupted due to salt & pepper noise only. We
want to apply 5x5 alpha trimmed mean filter to restore the noisy pixel at the center
255 12 255 0 15
255 14 13 0 0
16 12 255 14 255
16 13 13 0 255
255 0 10 0 18

What minimum value of d is needed to completely ignore the influence of noise in the mean
when restoring the centering noisy pixel.
43. In order to remove the salt and pepper noise, a two-step process that first detect salt and
pepper noise followed by application of a certain type of filter to the detected noisy
pixels. The detection step is kept simple by considering all those pixels that either have
intensities of either 0 or L-1 as noisy. Is this detection method accurate? If not, what is
the limitation of the discussed detection method?
44. Describe how median filter provides better edge preservation when compared to the mean
filter. Provide examples of responses of 1x5 median and mean filter on ramp and step
edges.

45. Write a python function named selectiveMean that apply 3x3 mean filter to the pixels
having intensities in the range from 100-135 (100 and 135 are included). You may use
predefined python functions and libraries.
46. Write a python function named selectiveMedian that apply 3x3 median filter to the pixels
having intensities of 0 and 255. You may use predefined python functions and libraries.
47. Describe what type of noise the following filter suppress and what kind of noise produce
the case of failure. Provide the reason of your answer

48. Describe what type of noise the following filter suppress and what kind of noise produce
the case of failure. Provide the reason of your answer

49. Describe what type of noise the following filter suppress and what kind of noise produce
the case of failure. Provide the reason of your answer

50. Describe what type of noise the following filter suppress and what kind of noise produce
the case of failure. Provide the reason of your answer

51. A weighted median filter is a filter that takes median with the help of weight window.
One example of a 3x3 weighted window is

1 2 1
2 3 2
1 2 1
It means when applied, each corresponding pixel will be repeated according to the weight
assigned. For example, if that 3x3 weighted window is applied to the following patch from input
image
25 30 19
20 23 17
21 22 35
Intensity 25 will be repeated once, 23 will be repeated thrice and so on before getting median
values. So the following will be the sorted pixels
17, 17, 19, 20, 20, 21, 22, 22, 23, 23, 23, 25, 30, 30, 35
Hence median will be 22 in this case. A 5x5 weighted window will look something like
1 2 3 2 1
2 4 5 4 2
3 5 7 5 3
2 4 5 4 2
1 2 3 2 1
Describe how weighted Median filter it is better when compared to the conventional median
filter?

52. A weighted median filter is a filter that takes median with the help of weight window.
One example of a 3x3 weighted window is

1 2 1
2 3 2
1 2 1
It means when applied, each corresponding pixel will be repeated according to the weight
assigned. For example, if that 3x3 weighted window is applied to the following patch from input
image
25 30 19
20 23 17
21 22 35
Intensity 25 will be repeated once, 23 will be repeated thrice and so on before getting median
values. So the following will be the sorted pixels
17, 17, 19, 20, 20, 21, 22, 22, 23, 23, 23, 25, 30, 30, 35
Hence median will be 22 in this case. A 5x5 weighted window will look something like
1 2 3 2 1
2 4 5 4 2
3 5 7 5 3
2 4 5 4 2
1 2 3 2 1
Center weighted median filter, is derived from weighted median filter, however, it
assigns weights of 1 to all the pixels except for the center. An example of weights of
center weighted median filter can be the following

1 1 1
1 3 1
1 1 1
It means if such center weighted median filter is applied to a pixel with a 3x3
neighborhood
25 30 19
20 23 17
21 22 35
The sorted pixel will have 23 repeated 3 times, however, others will be repeated once
only

17, 19, 20, 21, 22, 23, 23, 23, 25, 30, 35
Hence median will be 23.
What will happen if we set all weights of weighted median filter or center weighted
median filter to 1? Also, implement center weighted median filter using python.

Morphological Image Processing

53. If skeleton operation is applied to binary image, the output gives you a thin line in
presence of some unwanted branches as can be seen from the regions highlighted using
the circle. Name a technique that remove these branches.

54. Image on the right is the result of applying erosion twice to the input binary image on the
left. As you can see that the noisy lines are gone, however, it resulted in making the text
thin. What approach should be followed in order to get back the actual thickness.
55. Write down the steps to obtain the output image on the right by processing the input
image on the left (Squares in the output image has the width of equal that or more than 13
pixels).

56. What morphological operation can be performed in order to connected the low-
thresholded edge map to high-thresholded edge map in hysteresis thresholding.

57. What would you suggest to remove the noisy items from the image below
58. Write morphological operations required to convert the input image to the output image
59. Write morphological operations required to convert the input image to the output image

Image Enhancement
60. Histogram equalization on RGB image result in reducing efficiency 3x. What would you
suggest to apply histogram equalization to the color image with less complexity?

61. What is image normalization? How and why it is performed? Your answer should not
exceed eight lines.

62. Give a single intensity transformation function for spreading the intensities of image so
the lowest intensity is L/4 whereas the highest is 3L/4.

63. Suppose that you want to perform Deep Learning-based semantic segmentation to
segment the optic disk in a retinopathy image. However, in order to get improved
segmentation, you want to enhance the optic disk. Some of the examples of such images
are shown below having a round-shaped optic disc on the left side. Which operation will
you use and provide the reason of choosing such operation. Answer within five lines.

64. The equation for the gamma correction is

γ
c=c r
Describe within three lines why c=1 is mostly preferred? Also, describe a scenario
within eight lines where we need to choose the value of c >1.

65. Give a single intensity transformation function for spreading the intensities of image so
the lowest intensity is 0 whereas the highest is 3L/4.
66. Define what is unsharp masking? Provide and discuss its mathematical equation. In
which case, high boost filtering is performed in unsharp masking?
67. Briefly describe how unsharp masking sharpen the edges by applying unsharp masking to
the signal shown in figure below. Draw response of each step as well.

68. Briefly describe how unsharp masking sharpen the edges by applying unsharp masking
to the signal shown in the figure. Draw response of each step as well.

69. Given the image I and the intensity transformation T(I) shown in Figure, indicate which
image in the second row correspond to the transformed image. Please provide the reason
as well.
Image Segmentation
70. Use thresholding to segment Image I such that the resulted segmented image looks like
G?

The histogram of image I is as follows:

a- Describe the relation between image I and the histogram.

b- Roughly specify the threshold(s).

c- Describe an algorithm that thresholds image I to image G using the thresholds selected in step

71. Describe within three lines that why second order differential operator is highly prone to
the noise?
72. Categorize the following operations into intensity transformation and spatial filtering.
Provide a reason for your answer. You should answer within eight lines in total.

i. Histogram projection ii. Using sobel to detect corners

73. Suppose you have the following equation

θ=tan
( )
−1 G y
Gx

Where G y and G x are y and x-directional gradients respectively. Please discuss within five lines
where and how the parameter θ can be used and how?

74. Use Sobel operator to compute the gradient magnitude and direction at the center pixel in
the following image?

75. Figure i shows the original image whereas others show applications of some image
processing operation to the input original image. Identify the operations as well as
provide a two to four-lines explanation of each answer.
i ii iii

Iv v vi