fuzzycmeans.m

% This program illustrates the Fuzzy c-means segmentation of an image. 
% This program converts an input image into two segments using Fuzzy k-means
% algorithm. The output is stored as "fuzzysegmented.jpg" in the current directory.
% This program can be generalised to get "n" segments from  an image
% by means of slightly modifying the given code.
clear all;clc;
in=input('enter an image to segment','s');
IM=imread(in);
IM=double(IM);
figure(1)
imshow(uint8(IM))
[maxX,maxY]=size(IM);
IMM=cat(3,IM,IM);
%%%%%%%%%%%%%%%%
cc1=8;
cc2=250;

ttFcm=0;
while(ttFcm<15)
    ttFcm=ttFcm+1
    
    c1=repmat(cc1,maxX,maxY);
    c2=repmat(cc2,maxX,maxY);
    if ttFcm==1 
        test1=c1; test2=c2;
    end
    c=cat(3,c1,c2);
    
    ree=repmat(0.000001,maxX,maxY);
    ree1=cat(3,ree,ree);
    
    distance=IMM-c;
    distance=distance.*distance+ree1;
    
    daoShu=1./distance;
    
    daoShu2=daoShu(:,:,1)+daoShu(:,:,2);
    distance1=distance(:,:,1).*daoShu2;
    u1=1./distance1;
    distance2=distance(:,:,2).*daoShu2;
    u2=1./distance2;
      
    ccc1=sum(sum(u1.*u1.*IM))/sum(sum(u1.*u1));
    ccc2=sum(sum(u2.*u2.*IM))/sum(sum(u2.*u2));
   
    tmpMatrix=[abs(cc1-ccc1)/cc1,abs(cc2-ccc2)/cc2];
    pp=cat(3,u1,u2);
    
    for i=1:maxX
        for j=1:maxY
            if max(pp(i,j,:))==u1(i,j)
                IX2(i,j)=1;
           
            else
                IX2(i,j)=2;
            end
        end
    end
    %%%%%%%%%%%%%%%
   if max(tmpMatrix)<0.0001
         break;
  else
         cc1=ccc1;
         cc2=ccc2;
        
  end

 for i=1:maxX
       for j=1:maxY
            if IX2(i,j)==2
            IMMM(i,j)=254;
                 else
            IMMM(i,j)=8;
       end
    end
end
%%%%%%%%%%%%%%%%%%
  figure(2);
 
imshow(uint8(IMMM));
tostore=uint8(IMMM);
imwrite(tostore,'fuzzysegmented.jpg');
end

for i=1:maxX
    for j=1:maxY
         if IX2(i,j)==2
            IMMM(i,j)=200;
             else
             IMMM(i,j)=1;
    end
  end
end 

%%%%%%%%%%%%%%%%%%
IMMM=uint8(IMMM);
figure(3);
imshow(IMMM);
disp('The final cluster centers are');
ccc1
ccc2