CITS4402 Computer Vision

Week 6 Colour Based Segmentation

Due: Check the deadline on cssubmit.

Colour plays an important role in segmentation and object recognition. For example, the iris in the previous lab could have been easily segmented using its distinct colour from the rest of the image. In this lab, you are required to make a GUI that performs colour based segmentation on an input image.

Notice how the iris has been segmented in the left image and the green peppers in the right image. You should try segmentation in the RGB and HSV colour spaces both. Use two thresholds to select a narrow colour window.

Challenge [extra marks]: If you are looking for some challenge, modify your program so that it can perform human skin detection automatically i.e. no manual threshold adjustment. 

Submit your code with the best colour space option preselected. This is a relatively simple lab so you can complete it quickly and get to your project.

1. Your code should not give any kind of warnings or error messages
2. When the GUI starts, its should load the "peppers.png" image by default. This image comes with Matlab.
3. When the GUI starts, the axis should not appear on the right side image window, even though there is no image there.
4. When the Load Image button is pressed, it should show all types of files in the folder (not just .m files)
5. Your GUI window should be resizable.

Submission Requirements
- Submit your lab5.fig file
- Submit your lab.m file
- Sumit your favourite images

School of Computer Science and Software Engineering

This Page

Updated by: Ajmal Mian