IMAGES_BASICS

Contents

clear the workspace ...

... and update Matlab caches

clear all
rehash

image representation in Matlab

2- or 3-dimensional array

imrgb = imread('Kravy_na_podzim.jpg');
imgray = rgb2gray(imrgb);
whos('imrgb','imgray')

  Name         Size                           Bytes  Class

  imgray     480x551                         264480  uint8 array
  imrgb      480x551x3                       793440  uint8 array

Grand total is 1057920 elements using 1057920 bytes

Display images

figure(1); clf
imshow(imrgb);
title('RGB image')
axis on
figure(2); clf
imshow(imgray)
title('Gray scale image')
axis on

Data types pitfalls

For some image arithmetics it will be needed to change the default integer data type into double.

Matlab DOES recognize data types. Even if the values of the variables do not change the output (displaying and saving) could. Be aware of it! This caused many confusions in past.

for type uint8 Matlab expects values

and for type double it expects

imrgb_float = double(imrgb);
imgray_float = double(imgray);
figure(3); clf
subplot(2,2,1); subimage(imrgb);
title('RGB image with integers');
subplot(2,2,2); subimage(imrgb_float);
title('RGB image with floats');
subplot(2,2,3); subimage(imgray);
title('gray image with integers');
subplot(2,2,4); subimage(imgray_float);
title('gray image with floats');

Saving images and figures

% saving bitmaps
imwrite(imgray,'examples_of_saving.png');

% saving (printing) graphics (figures)
% activate the proper figure
figure(3);
% save as png bitmap
print -dpng figure3.png
% or as function
% save as color encapsulated postscript
% eps file may get very big. However, it is very useful for publishing.
print('-depsc','figure3.eps')

Indexing of gray images

is pretty straightforward

% Subimage by specifing coordinates
imgray_sub = imgray(300:400,50:350);
% find all pixels where the intensity exceeds threshold
threshold = 128;
idx = imgray<threshold;
% indexing by a logical conditions
imgray_changed = imgray;
imgray_changed(idx) = 0;

figure(4); clf; imshow(imgray_sub); axis on;
figure(5); clf; imshow(imgray_changed); axis on;

Indexing of RGB data

indexing of multidimensional data is slightly more complicated

% Subimage by specifing coordinates
imrgb_sub = imrgb(300:400,50:350,:);
% find all pixels where at least one from RGB values is lower threshold
threshold = 70;
idx = any(imrgb<threshold,3);
% create the index matrix for indexing RGB image
idxmat = repmat(idx,[1,1,3]);
% indexing by a logical conditions
imrgb_changed = imrgb;
imrgb_changed(idxmat) = 0;
% and more straighforward replacing of all pixels where
% the RGB average exceeds 200 (check the sky and the rock)
imrgb_changed(repmat(mean(imrgb,3)>200,[1,1,3])) = 255;

figure(6); clf; imshow(imrgb_sub); axis on;
figure(7); clf; imshow(imrgb_changed); axis on;

Published with MATLAB® 7.0.1