Edge detection in colour images

COMPSCI 775ST Assignment 1
From: 31/07/2002
Due: 09/08/2002 2:00pm

• Aim

• Resource

• Goal

• Details

• Questions to be answered

• Example of results

• Understand edge detection in colour images.
• Know more about colour images.
• Understand local window operation.

There are a lot of image files available. Choose one or a few of them that are appropriate for the assignment. The followings are some images you may use in the assignment.

• brick.0000.jpg
• brick.0004.jpg
• buildings.0003.jpg
• fj305.jpg
• gbmus7.jpg
• wood.0002.jpg

You need to investigate, comment and compare the performance of at least TWO edge detection methods of your choice.

Edge detection method	Comments
SOBEL operators	Simple, briefly mentioned in the lecture.
Laplacian-of-Gaussian (LoG) filter	Mentioned in the lecture.
Castan edge operators, Gradient method	Discussed in the lecture in detail.
Castan edge operators, Zero-crossings method	Discussed in the lecture in detail.
PSEUDO-LAPLACE operators	Simple, outside of the scope of this paper.
Morphological Edge operator	Simple, outside of the scope of this paper.
Difference-of-Gaussian (DoG) filter	Outside of the scope of this paper.
KIRSCH operators	Outside of the scope of this paper.
Deriche Edge Operator	Outside of the scope of this paper.
Contra-harmonic Filter	Outside of the scope of this paper.

• Choose one or a few appropriate colour image(s).
• Describe the image format and why you have selected this format.
• Describe and implement the edge detection methods that you have chosen.
• Apply the edge detection methods to suitable images.
• You may vary the local window size and comment on the difference.

• What are the merits and demerits of the selected edge detection methods?
• Under what conditions do you expect the selected edge detection method to better perform?
• Discuss the colour models the you have used and why you have selected them.
• With the chosen edge detection method, display and comment the obtained results.


Compare the results to your expectations of the chosen edge operators.