ECE 563 Assignment 6

Images and m-files for this assignment may be downloaded from impro6.zip.

1. You are given three "mystery" aerial images, obviously of the same location, but with differing brightness variations. These images might be three spectral bands, or the same image processed by some algorithm. Act on the assumption that the images are three spectral bands, decide which image should be red, green, and blue and produce a composite 24-bit image.

   

2. Take your scanned photograph from previous assignments and display its components as shown in these examples.

3. Do the following:
   1. Use rgb2hsv and hsv2rgb to rotate the hues of your scanned color photograph by 120-degrees (both directions).
   2. Change your scanned image from RGB to BRG. That is display red as green, green as blue, and blue as red. Compare the resulting false color image to those produced in the previous part.

4. Find matrices that convert XYI = M RGB and RGB = M^-1 XYI, where

    
   X = R -(G+B)/2;
   Y = sqrt(3)/2*(G-B);
   I = (R+G+B)/3;
   

   Convert your color photograph to XYI coordinates. Plot the co-histogram of the XY coordinates. Finally set each of X and Y to zero, in turn, convert back to RGB and show the resulting image. Explain what you see.

5. Convert the following images of fruit to XYI, plot the co-histogram (and grayscale histogram). Now pick a characteristic color (perhaps by averaging the colors in a small region) of an apple and an orange. Generate grayscale images for the Euclidian distance and Mahalanobis distance from those characteristic colors. Find the best threshold that selects the regions most "apple-like" or "orange-like".

   fruit2441.jpg	fruit2442.jpg

6. Given the program colorseg from DIPUM, apply the algorithm to the problem of distinguishing apples and oranges in the previous exercise.

Maintained by John Loomis, last updated 8 March 2016