Scribd
Upload
54 views

Ass3 2022

This document provides instructions for Assignment 3 in COMP 478/6771 Image Processing at Concordia University. It includes two parts: theoretical questions about image processing concepts like spatial masks and the Radon transform, and programming questions implementing edge detection algorithms like the Laplacian of Gaussian and Canny edge detectors on an image. Students are asked to prove properties of the Radon transform, discuss which phase combination better reconstructs an image, and compare edge detection results using the two algorithms.

Uploaded by

saumya11599
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
54 views

Ass3 2022

This document provides instructions for Assignment 3 in COMP 478/6771 Image Processing at Concordia University. It includes two parts: theoretical questions about image processing concepts like spatial masks and the Radon transform, and programming questions implementing edge detection algorithms like the Laplacian of Gaussian and Canny edge detectors on an image. Students are asked to prove properties of the Radon transform, discuss which phase combination better reconstructs an image, and compare edge detection results using the two algorithms.

Uploaded by

saumya11599
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Concordia University

Department of Computer Science & Software Engineering


COMP 478/6771 Image Processing
Assignment 3 - Due Date: Nov 15, 2022

Part I: Theoretical questions

1. (10 points) Given a 3x3 spatial mask that averages the four closest neighbors of a point (x, y)
but excludes the point itself from the average (the central point). Find the equivalent filter
H(u,v) in the frequency domain. For this question, please use Property 3 of Table 4.4 in the
textbook, which summarizes the property of spatial translation for an image:

2. (16 points) Prove the validity of the following properties of the Radon transform:

(a) (8 points) Linearity: The Radon transform is a linear operator (use definition of linearity).
(b) (8 points) Translation property: The radon transform of f (x -x0 , y -y0 ) is
g(𝜌 -x0 cos 𝜃 - y0 sin 𝜃, 𝜃).

Part II: Programming questions (24 points)

1. (6 points) Let IA , IB be two completely different gray level images, and FA , FB be the Fourier
transforms of IA and IB respectively. The Fourier transform FA of IA consists of its magnitude |
FA | and phase Ω(FA ). Similarly, the Fourier transform FB of IB consists of the magnitude | FB |
and phase Ω(FB ). We want to reconstruct the image IA in the following experiment: a new
image I1 is obtained by inverse Fourier transform using the of magnitude | FA | of image A and
phase Ω(FB ) of image B. Similarly, a new image I2 is obtained by inverse Fourier transform
using the magnitude | FB | of image B and phase Ω(FA )of image A. In your opinion, which one
of the two reconstructed images I1 and I2 is the better reconstruction of the original image IA .

Please use the two images in the assignment package to justify for your answer.
2. (18 points) Download the image “house.tif” from the assignment package then perform edge
detection using existing MATLAB functions (with the parameter choices of your own) for:
a) Laplacian of Gaussian (Marr-Hildreth) edge detector
b) Canny edge detector

1) (4 points) Briefly list the steps involved in implementing the edge detectors.
2) (4 points) Explain how edge linking (the final step of the Canny algorithm) was
implemented. Does the first method need this step?
3) (4 points) List the parameters that determine the performance of the algorithms. What
parameter values did you use and why?
4) (6 points) Show and compare the results obtained by the two methods (give some
comments).

You might also like