Colour Vision Exercise for the Colour Vision Lecture

This exercise is on colour in image analysis
  1. Download ColourMM.zip and extract it with winzip.
  2. Discuss the differences between the colour spaces, RGB, HSV. May be review the respective slides from the lecture.
  3. start matlab and try to convert an RGB image to HSV space
    1. Go to directory ColourMM/Images. (You can get help by typing help command)
    2. I=imread('dP940.tif');
    3. imshow(I)
    4. display only e.g. the red component : imshow(I(:,:,1))
    5. use rgb2hsv (help rgb2hsv)
    6. display the different components of the hsv and discuss them
  4. go to the directory ColourMM, start matlab and type ColourMM. You will get a user interface. Some of the function are already working and some you will have to implement.
    1. Modelling of the chromaticity of the reflecting material: You can select a light source (927 = 2700K, 940 = 4000K, 950 = 5000K, 965 = 6500K) and a reflecting material (skin, blue-, green-, red-plastic). If you press calculate and plot chromaticities you will get a '*' indicating the light sources chromaticity. Your task is now to implement the calculation and visualization of the materials reflections in the file exercise1.m. Have a look at the slides around spectral integration and independence to brightness if necessary.
    2. By pressing on Load Image you can load an image. The number in the file name indicates the light source used for taking the image (927 = 2700K, 940 = 4000K, 950 = 5000K, 965 = 6500K). Load an image (for example Gundega940.tif), press Select Area and show chromaticities. You get a new figure with the image. Now you can select an area (take the forehead) and doubleclick to stop the selection. The chromaticities of the selected pixels will be displayed after some calculation time (matlab is not very fast) in the chromaticity plane. Compare the modelled with the measured data.
    3. Implement a simple segmentation. The file exercise2.m gives a framework for this. We will discuss possibilities together. (Images starting with x are downsampled which makes the processing faster).