Hi,I'm doing my project about image processing, using the co-occurrence matrix but I don't know the algorithm of the co-occurrence(glcm) matrix can any one help me to get this algorithm and to get results like this. thank you a lot

Answer 1

[Walter Roberson]

http://www.mathworks.com/help/images/ref/graycomatrix.html#bu5tkok-5

Answer 2

[Image Analyst]

It basically tells you how often each gray level occurs next to each of the other gray levels.

[rouh 238]

I use this function to calculate the glcm of each image but I don't get good result

function [t]= glcm(I)
glcmm = graycomatrix(I,'Offset',[0 1, -1 1, -1 0, -1 -1]);
stats = graycoprops(glcmm,{'contrast','homogeneity','Correlation','Energy'})
t = struct2array(stats);

[Walter Roberson]

If you are getting an error message, you need to post it.

If the values are not useful for your purpose, then it could be that you need to pass different Offset or different options.

[rouh 238]

No I don't get any error message but when I calculate the distance between the query image and the other images I get the same result for each query image. I use this formula to calculate the distance:

   valeur=sqrt(abs(m1(1)-data{K}(1))+abs(m1(2)-data{K}(2))+abs(m1(3)-data{K}(3))+abs(m1(4)-data{K}(4)));

[Walter Roberson]

It is not obvious that that calculation has anything to do with glcm .

What is class(m1), class(data{K}) ?

[rouh 238]

this is my code

I = rgb2gray(a);
t1=glcm(I);
j=1;
projectdir = 'img\texture';
list = dir(projectdir);
list([list.isdir]) = []; 
filenames = natsort( {list.name} );
for K = 1 : length(filenames)
  ch = filenames{K};
  nomImage = fullfile(projectdir, ch);
  storedStructure = load (nomImage);
  data{K} = storedStructure.t;
  valeur=sqrt(abs(m1(1)-data{K}(1))+abs(m1(2)-data{K}(2))+abs(m1(3)-data{K}(3))+abs(m1(4)-data{K}(4)));
  resulat(j)=valeur;
        j=j+1;
    end
    resultatFinal=sort(resulat,'descend')
    for k=1:j-1
        X=find(resultatFinal(k)==resulat);
        indice(k)=X(1);
    end
   n=2;
    for k=1:12
        nomImage=strcat('image\',num2str(indice(k)),'.jpg');
        I=imread(nomImage);
        axes(handles.(sprintf('axes%d', n)));
        n=n+1;
         imshow(I);
    end
  end

[Walter Roberson]

You do not show any definition for m1, but whatever it is, it is not changing with the results of glcm .

We cannot tell from your code what class(m1) or class(storedStructure.t) are.

[rouh 238]

I'm sorry it's not this formula, this is mine :

    valeur=sum(abs((data{K})- t1));

[rouh 238]

(storedStructure.t): it's the composite where we find the glcm of the images of the data base

   for k=0:999
  a=imread(sprintf('image\\%d.jpg',k));
 I1 = rgb2gray(a);
 t=glcm(I1);
  save(sprintf('img\\texture\\%d.mat',k),'t');
 end