Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Practice Questions

1. a) An image contains number of boxes. Indicate a Sobel filter to detect Horizontal edges
of boxes.
b) How would the following transformation affect the appearance of an image? Draw the
curve and explain.
s = 100 log10( 1 + r)
c) How is a Gaussian filter better than the simple mean filter?
d) An image is found to be corrupted with salt and pepper noise. What filter would you
suggest to remove the effect of the noise. Justify your reason.
e) Why is it preferred to represent lines in Hough Space using polar coordinates rather than
cartesian coordinates?
f) Comment on the following: An image A is eroded by a 5x5 structuring element B, followed
by Dilation using B. The final image would be same if A was first dilated by B and then
eroded by B. Justify your answer briefly.
g) Indicate the morphological operation to obtain the boundary BN of an object from a
binary image M.
h) An image contains 3 vertical lines and 2 horizontal lines. Indicate how many major
intersections will be there in the corresponding Hough Space. Justify your answer.
i) In the class we have seen how Lucas Kanade method can be used to determine motion of
an object with a fixed camera. How to determine motion if the scene is stationary but the
camera is moving?

j) Indicate if following Lucas Kanade formulation is correct. If not make the corrections on it.

Σ I x Ix Σ Ix I y u Σ I t It
=
Σ Ix I y Σ It Iy v Σ I x It
k) The DFT of an image is of size 8x8 contains 25 in the top left position, while the remaining
elements are all zeroes. What can you say about the size and type of original image.

l) Work out the number of HOG parameters for an image of size 64x64.

m) Why do you get two images, namely, magnitude spectrum and phase spectrum when you
apply DFT on an image?
2. Apply 3x3 Laplace filter on the following image and work out the outputs when the filter
is centered on gray values marked 33 and 82. What relevant information do the output
values signify?

30 30 30 30 30 30
30 31 30 33 80 30
30 30 80 80 80 80
30 36 80 82 80 80
30 30 80 80 80 80

3. Work out the average entropy of the following image segment

40 95 95 95 40 95 95 95
40 95 95 95 40 95 95 95
40 95 95 95 40 95 95 95
40 95 95 95 40 95 95 95
40 95 95 95 40 95 95 95
40 95 95 95 40 95 95 95

4. Draw a rough plot of histogram of following image with 8 distinct gray levels.

Gray 0 1 2 3 4 5 6 7
Level
Number 420 672 560 420 280 224 140 84
of Pixels

Carry out histogram equalization on this image. Draw the new histogram and comparing the
two histograms comment on the relative quality of the images. Figure out if the picture is
going to be brighter or darker after equalization.
5. An image contains four circles. The size of the image is 128x128. The circles are of radius 20.
Develop the Hough parameter space equation to detect the circles. Draw the Hough Space grid and
indicate the axes. How many major intersections?

6. An image contains three circles . The size of the image is 128x128. The centres of the circles are
located on the line x = 60. Develop the Hough parameter space equation to detect the circles. Draw
the Hough Space grid and indicate the axes. How many major intersections?

7. The histogram of an image with 256 gray levels is shown below. Work out the
compression ratio possible with a Lossless scheme.

Level 80 110 135 160 200 240


pixels 30 50 120 180 220 400
8. Indicate a scheme for removing small noisy patches from binary image and keep all
objects greater than 15x15
9. Two images are shown below. Using the 4x4 patch from top left corner work out the – 450
GLCM matrices. Use energy and contrast scalar descriptive measures to show that the two
image textures are different.
28 28 28 4 28 28 28 4 28 28
4 28 4 4 4 28 4 4 4 28
4 28 28 4 4 28 28 4 4 28
4 4 4 4 4 4 4 4 4 4
28 28 28 4 28 28 28 4 28 28
4 28 4 4 4 28 4 4 4 28
4 28 28 4 4 28 28 4 4 28
4 4 4 4 4 4 4 4 4 4

28 28 28 28 28 28 28 28 28 28
4 4 4 4 4 4 4 4 4 4
28 4 28 4 28 4 28 4 28 4
4 4 4 4 4 4 4 4 4 4
28 28 28 28 28 28 28 28 28 28
4 4 4 4 4 4 4 4 4 4
28 4 28 4 28 4 28 4 28 4
4 4 4 4 4 4 4 4 4 4

10. Based on the knowledge where a point of known intensity moves between two frames of a
video we can get the Optical Flow Equation. Indicate why we cannot directly determine the
optical flow components from this equation? Indicate a simple way to resolve this problem.
11. Draw a pictorial representation of an ideal low pass filter that can be applied to a gray scale
image. Indicate why it is not really “ideal” for practical applications.
12. Indicate the difference between face image and eigenface.
13.
Region filling is to be carried out on following image using Morphological operators. Pixels
not marked in the image have value 0. Let the initial array X0 consist of just pixel P with
value 1. In the “final Result” section below, indicate which pixels inside the region get value
1 in iteration X2. Use the structuring element
010
111
010

1
2 1 1 1 1 1
3 1 1
4 1 P 1
5 1 1
6 1 1
7 1 1 1 1 1 1 1
8

1
2
3
4
5
6
7
8

Final Result
1
2
3
4
5
6
7
8

14. Given f as vector [ 12 20 16 8] and N as 4 compute Fourier Coefficient F(3)

You might also like