ising2.m

MRFImageSegmentationDemoWithMatlab from Gatech

%Realization of Binary Markov Random Field  by Metropolis
 clear all
 close all
%-----------------figure defaults
disp('Ising by Metropolis')
lw = 2;  
set(0, 'DefaultAxesFontSize', 16);
fs = 14;
msize = 5;
randn('state',3) %set the seeds (state) to have 
rand ('state',3) %the constancy of results

pixelX = 256;
pixelY = 256;
 J = 0.85;
 el = 0;
%load klaus256; F=  (2.* (ima256 > 0.4) - 1);
%F = -ones(Ypix,Xpix);
F = ( 2 .* ( rand( pixelX, pixelY ) > 0.5 ) - 1 );

figure(1)
colormap(gray)
imagesc(F)


for jj = 1:500
%while 1,
for k = 1 : 10000
% Select a pixel at random
ix = ceil( pixelX * rand(1) ); 
iy = ceil( pixelY * rand(1) );
Fc = F( iy, ix ); %the value at position (ix, iy) 
pos = ( ix - 1 ) * pixelY + iy; %  univariate index of pixel
neighborhood = pos + [-1 1 -pixelY pixelY]; % Find indicies of neighbours
neighborhood( find( [iy == 1    iy == pixelY    ix == 1   ix == pixelX] ) ) = []; 
% pesky boundaries...thrash them
  nagree =  sum( Fc == F(neighborhood) ); 
  ndisagree = sum( Fc ~= F(neighborhood) );
change = nagree - ndisagree;
if rand(1) < exp( -2 * J * change ) % if change<0, proposal is taken wp 1
    F( iy, ix ) = -Fc;                %accept proposal 
end
el = el + 1;
end
figure(2); imagesc(63*F); colormap(gray); title(['Iteration #  ' num2str(el)]);
drawnow
end
figure(2); imagesc(63*F); colormap(gray); title(['Iteration #  ' num2str(el)]);
%print -depsc 'C:\Brani\Courses\Bayes\Handouts\Working16\Figs\metroising1_85.eps' %jj=10
%print -depsc 'C:\Brani\Courses\Bayes\Handouts\Working16\Figs\metroising2_85.eps' %jj=50
%print -depsc 'C:\Brani\Courses\Bayes\Handouts\Working16\Figs\metroising3_85.eps' %jj=500