Question 1-Canny Edge Detector (10 Points) : Fundamentals of Computer Vision - Midterm Exam Dr. B. Nasihatkon

Fundamentals of Computer Dr. B.
Vision - Midterm Exam Nasihatkon
Name: ID: Ordibehesht 1396 - May 2017
Question 1- Canny Edge detector (10 points)

Consider the images below. The leftmost image represents the result of Canny low- and
high- thresholding. The black lines represent the strong edges and the gray lines indicate
the weak edges weak edges. Remember, weak edges are edges whose gradient
response (magnitude) is between the low and the high thresholds, and strong edges are
those whose gradient magnitude is larger than the high threshold. On the outline image on
the right draw the final edges obtained by the Canny edge detector. If you screw up one of
the outline images, use the other one, but, only use one of them (10 points).
(Cross out the image to the

left if you are using this.)
K. N. Toosi University of Technology

Question 2: Correlation, Convolution and median filtering (18 points)
Consider the filter and the image below.
Filter Image
0 0 1 3 2
-1 0 2
2 3 6 0 3
0 2 1
5 4 2 5 7
-3 1 2
1 2 0 4 4
A) What is the output of a correlation operation on the image using the above filter?
What is the result of convolution? Fill in the blank cells below (12 points.)
Correlation Convolution
X X X X X X X X X X
X X X X
X X X X
X X X X X X X X X X
B) Apply a 3 by 3 median filter to the image above and print the results in the blank
cells below (6 points.)
X X X X X
X X
X X
X X X X X

Question 3 - Gradient / Edge detection (18 points)
Sobel Filters (for correlation) Image
0 0 1 3 2
-1 0 1 -1 -2 -1
2 3 6 0 3
-2 0 2 0 0 0
5 4 2 5 7
-1 0 1 1 2 1
1 2 0 4 4
dx dy
A) For the image I above compute the gradient (dI/dx, dI/dy), magnitude of gradient
and the angle of gradient for pixel locations A, B and C marked below. You have
to use the Sobel filters (above) to compute the derivatives (no normalization is
needed). You need to write down how you derived each of these quantities. You
may use arctan to represent the angles. The y axis is pointing downwards. (15
points)
B C
B) Draw the gradient direction at each of the points A, B and C (3 points).

Question 4 - Histogram (18 points)
Look at the following histograms for images A, B, C, D, E and F, all images of the same
scene. Zero and 255 represent black and white intensities respectively.
A B C
D E F
A) Which of the above histograms correspond to the following

image? Explain your answer. (6 points)
B) Which of the images D and E has a higher contrast? Why? (4 points)
C) How does image B look like compared to image A? Explain. (4 points)
D) What is the relation between images C and D? Explain your answer. (4 points)

Question 5 - Hough Space (18 points)
Assume that the lines are parameterized with an angle θ of
the line normal and a distance ρ from the origin:
A) For each of the lines in the image space below (left), draw the corresponding point
in the hough space (right). Use letters A, B, C and D to mark the points. You need
to also write the point coordinates ( arctan(2) ≃ 64 degrees). Assume ρ ≥ 0 and
0 ≤ θ < 180. (12 points)
Image Space Hough Space
B) For each of the points in the hough space below (left), draw the equivalent line in the
image space (right). On each line draw at least two points (for example, intersection with
the x and y axes, if possible. Or two other points). You just need to draw a line segment in
the first quadrant of the coordinate system ( x ≥ 0, y ≥ 0) . (6 points)

Hough Space Image Space
Question 6 - Harris Corner Detection (18 points)

Assume that we have already computed the gradient (dI / dx, dI / dy) for an image. The
two arrays below contain I x = dI / dx and I y = dI / dy at each pixel of the image.
I x = dI / dx I y = dI /dy
3 2 1 -1 -1 2 3 1 1 -1
4 3 2 0 -1 2 3 2 -1 1
4 3 4 2 1 2 4 4 1 2
1 1 3 2 2 -1 0 3 2 3
A) Compute the Harris matrix for the 3 by 3 window highlighted in the above images.
The Harris matrix is defined as (10 points)
where W is the window highlighted in the gradient images.
B) Compute the cornerness score for the above matrix C = det(H) − k trace(H)2 for the
above window, where det and trace represent the determinant and trace (sum of
diagonal elements) of a matrix. Use k = 0.04. (3 points)
C) What do you think we have in the above window? A corner? An edge? Or a flat
region? Explain your answer. (5 points)

Question 1-Canny Edge Detector (10 Points) : Fundamentals of Computer Vision - Midterm Exam Dr. B. Nasihatkon

Uploaded by

Copyright:

Available Formats

Question 1-Canny Edge Detector (10 Points) : Fundamentals of Computer Vision - Midterm Exam Dr. B. Nasihatkon

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Question 1-Canny Edge Detector (10 Points) : Fundamentals of Computer Vision - Midterm Exam Dr. B. Nasihatkon

Uploaded by

Copyright:

Available Formats

Fundamentals of Computer Dr. B.

Vision - Midterm Exam Nasihatkon

Name: ID: Ordibehesht 1396 - May 2017

Question 1- Canny Edge detector (10 points)

(Cross out the image to the

K. N. Toosi University of Technology

K. N. Toosi University of Technology

Sobel Filters (for correlation) Image

B) Draw the gradient direction at each of the points A, B and C (3 points).

K. N. Toosi University of Technology

A) Which of the above histograms correspond to the following

K. N. Toosi University of Technology

Image Space Hough Space

K. N. Toosi University of Technology

Question 6 - Harris Corner Detection (18 points)

where W is the window highlighted in the gradient images.

K. N. Toosi University of Technology

You might also like