Basic intensity transformations

Contents

clear everything

clear all

read input image

% orig.im = imread('wom.jpg');
% bigger image to emphasize the effect
orig.im = imread('wom.jpg');

histogram of the original image

orig.h=hist(double(orig.im(:)),0:255)./prod(size(orig.im));

cumulative distributive function

orig.cumh = cumsum(orig.h);
showdata(orig,1,'original image');

making negative: computation with and without loops, effect of memory pre-allocation

Check the elapsed time for computation. The Matlab is very effective in matrix computation and very ineffective when looping without memory pre-allocation

lut = 255:-1:0;
t = cputime;
neg.im = uint8(lut(double(orig.im)+1));
elapsed.lut = cputime-t;

looping with memory pre-allocation

neg.im = zeros(size(orig.im));
t = cputime;
for i=1:size(orig.im,1),
  for j=1:size(orig.im,2),
	neg.im(i,j) = lut(double(orig.im(i,j))+1);
  end
end
elapsed.allocated = cputime-t;

looping without memory pre-allocation

t = cputime;
for i=1:size(orig.im,1),
  for j=1:size(orig.im,2),
	negative(i,j) = lut(double(orig.im(i,j))+1);
  end
end
elapsed.nonallocated = cputime-t;

display elapsed times

types = fieldnames(elapsed);
for i=1:size(types,1),
    eval(['elapsed_time = elapsed.',types{i},';']);
    disp(sprintf('Elapsed time for %s: %f secs',types{i},elapsed_time));
end

Elapsed time for lut: 0.010000 secs
Elapsed time for allocated: 1.050000 secs
Elapsed time for nonallocated: 1.170000 secs

showdata(neg,2,'negative');

histogram equalization by the function from ipt

ekv2.im = histeq(orig.im);
showdata(ekv2,3,'equalized image by histeq');

Published with MATLAB® 7.0.1