How to create a .mat file to store multiple images and its histograms.

Hi,

I'm new to matlab and would greatly appreciate if anyone could guide me about .MAT file. I'm doing a image retrieval system which requires a database of 1000 images. Currently, i have a .mat file which contains the training data for 100 images but i would like it to increase 1000 images. How do i edit the .mat file or how to do i go about creating a .MAT file which it can store 1000 images and its histograms. Your help is greatly appreciated. Thanks.

Answer 1

[Walter Roberson]

For example,

for K = 1 : 1000
  filename = sprintf('image_%04d.tif', K);
  I = imread(filename);
  IGray = rgb2gray(I);
  H = hist(Igray(:), 32);  %32 bins in the histogram
  imgs(K).image = I;
  imgs(K).histogram = H;
end
save('ImageDatabase.mat', 'imgs');

[Walter Roberson]

image_%04d.tif is the pattern of the names of the images that are read in in this example, The pattern is 'image_' followed by a 4-digit number with leading 0's, followed by '.tif'

The example is a demonstration of how you might build up the big database by reading in a bunch of images that are currently in separate files, the names of which follow a numeric pattern. If you want to get file names from the directory, please see this FAQ

The number of bins to use for your histogram depends entirely on your purposes. Different investigations have found different number of bins to be better.

Note that in the sample code, I transformed the image to grayscale and did a histogram on the grayscale values (independent of position.) If you want to do RGB or HSV histograms, you would not do the rgb2gray(), and you would need to figure out how you want to create a histogram when you have 3 different color planes, R vs G vs B, or H vs S vs V would require 3 dimensional outputs... and then what?

[Revathi]

Hi, Thanks for the prompt reply. May i know, is this the specific image name (image_%04d.tif)? I'm actually working on a RGB and a HSV color model. How many bins should i declare for both? Thanks so much for your help.

[Fattah Tuhin]

Nice . This helped me lot. Thank you