1

I would like to analyze the roughness of skin for Honeydew and also Pear(Gong Pear) in images by using Gray Level Cooccurrence Matrix in MATLAB. I have converted the 2 pictures (1st picture contains Honeydew and 2nd picture contains the Pear) into grayscale and calculated GLCM for each of the picture in MATLAB by using graycomatrix. I have then obtained the 4 GLCM features (contrast, energy, homogeneity, correlation) by using graycoprops function in MATLAB. Both images for Honeydew and Pear are captured on a black background.

It is expected that the skin of Honeydew is rough and the skin of Pear is smooth. However, from the 4 GLCM features that I have obtained, I found that the values of GLCM features obtained for Honeydew and Pear are quite close to each other (ie: correlation for Honeydew is 0.8197 while for Pear is 0.8819).

How can I classify the roughness of skin based on the 4 GLCM features? Or is there any way to make the values of GLCM features more distinct?

Tonechas
  • 13,398
  • 16
  • 46
  • 80
user3809831
  • 11
  • 2
  • Have you tried using entropy? I have used it before to evalueate roughness http://www.mathworks.co.uk/help/images/ref/entropy.html – Ander Biguri Oct 13 '14 at 10:13
  • @AnderBiguri: Do you mean determine entropy of the created GLCM or straight away determine the entropy of the grayscale image? – user3809831 Oct 13 '14 at 10:24
  • 1
    I did it directly in the image. You may want to have a look to this Matlab blog post: http://www.mathworks.co.uk/help/images/examples/texture-segmentation-using-texture-filters.html – Ander Biguri Oct 13 '14 at 10:53

0 Answers0