The images ‘zebra_1.tif’, ‘zebra_2.tif’ … ‘zebra_6.tif’ (which can be found here), contain some fine specimens of a particularly textured herbivore. We will try to implement a zebra-detector by analyzing texture.

Task 1

First, try to implement your own GLCM-function that takes as an input an image window and the number of image grayscales and outputs a co-occurance matrix. Derive variance, contrast and entropy from the GLCM of a sliding windown of a suitable size.

Task 2

Try to use a simple thresholding of these features to mask out the zebras in the images.

Task 3

Compare your result with the first order texture measurements: variance and entropy by using built-in functions from some suitable library.

Task 4

If you have time, try to use Laws’ texture masks to analyze the image with a suitable mask or two. Remember to average energy over windows.

Happy zebra hunting.