Getting the maximum value of several matrices loaded from image files.

I have troubles with the max command applied to images files.

I have N=108 image files tif. Each file corresponds to a matrix. I need to load all the matrix/images and compute a matrix containing the maximum values among the 108 matrices.

My script to read the N=108 files is as follows:

imagelist = dir('*.tif');
N = numel(imagelist);
imdata = cell(1, numel(imagelist));
for k = 1: N
    imdata{k} = imread(imagelist(k).name);
end

I don’t know to compute the “maximum matrix” since imdata contains matrices as cells.

Using

   M=max(imdata{k})

indeed I obtain a row vector containing only the maximum values of the first cell.

Answer 1

[Image Analyst]

Something like this (untested):

maxImage = zeros(rows, columns, 'uint8');
overallMax = -inf;
for k = 1: N
    thisImage = imread(imagelist(k).name);
    % Get max of this image.
    thisMax = max(thisImage(:));
    if thisMax > overallMax
         overallMax = thisMax;
    end

      % Get max image, in case you really meant that.  Delete if not.
      maxImage = max(maxImage, thisImage);
  end

Of course you need to make sure that the dimensions (rows, columns, number of color channels) of all the images match up.

[Chiara]

Thanks, using the script above I get the Error due to using max command:”Mixed inputs must either be single and double, or integer and scalar double. All other combinations are no longer allowed.” Anyway the image have all the same rows and columns.

[Image Analyst]

You did not give the complete error message! What was this for? The max() function??? It does work though, as I demonstrated in this code: http://www.mathworks.com/matlabcentral/answers/105939#answer_115077

[Chiara]

The error refers to "uint8" command. I correct in order to have all values with the same precision (single). Now it perfectly works. I meant to have the max image though. Thank you a lot

[Image Analyst]

OK good. MATLAB has a rule that integers must be combined with integers of the same class. Usually that comes into play when you're adding or multiplying them but I guess you found out that it also applies when you use the max function. There's no such rule for floating point numbers. Anyway, I'm glad you got it working and accepted the answer.

Answer 2

[Doug Hull]

Do not store the images in cells. Just load each in turn, get the maximum value. Compare this maximum value to the previously found maximum value. Keep the maximum of those two. Repeat until all have been done.

[Chiara]

Thanks for your answer. I cannot convert the cells into matrices though. I’ve tried to automatically upload the 108 tif files I have, using strcat function but doesn’t work. My files are “200301_monthly_adg_443_gsm_P90.tif” The only part varying is the first “200301”. I have already implemented strcat function but not with images.