How to calculate median matrix of many .tif files with MATLAB?

Hi

I have for example 15 .tif files, some pixels with "NaN". How to find a matrix of the median of the 15 tif files

Is it possible?

to be more precice

The following link to subsample of the data:

https://www.dropbox.com/scl/fo/bwb5bek75hv6d0hiyv0ey/h?rlkey=7xsqnc762ta8ezcj8e8x53qap&dl=0

input: 15 .tif files x m rows x n columns

output: so I want the out put be 1 median x m rows x n columns

Cheers

[cui,xingxing]

Can you please explain what you mean by "a matrix of the median of the 10 tif files", is it the median of the names of the 10 tif files you have given or is it the median of the pixel values read? I can't clearly understand what you mean.

And can you provide the 10 tif files? Upload them as attachments so that we can easily investigate your question!

[Walter Roberson]

Should we extract the numbers from the file name, and use them to sort the file names in numeric order?

Or is the order not important to you as long as you read them all in? Will they be the only 10 .tif files in the directory?

[Rafat]

Dear Walter:

"Or is the order not important to you as long as you read them all in?"

the order is not important

Will they be the only 10 .tif files in the directory?

I have not more than 120 tif files

Cheers

[Rafat]

Dear Cui;

And can you provide the 10 tif files?

I added them to the original qustion.

for example:

input: 15 .tif files x m rows x n columns

output: so I want the out put be 1 median x m rows x n columns

Cheers

[Voss]

Your images are not all the same size, i.e., some are 87x207 pixels, some are 88x207 pixels, some are 88x208 pixels, and some are 87x208 pixels.

unzip Test.zip

Warning: Cannot open file "" for reading.

files = dir('*.tif');
filenames = fullfile({files.folder},{files.name});
N = numel(files);
sz = zeros(N,2);
for ii = 1:N
    files(ii).data = imread(filenames{ii});
    sz(ii,:) = size(files(ii).data);
end
disp(sz)

    87   208
    87   207
    87   208
    88   207
    88   207
    87   208
    87   208
    88   207
    87   208
    87   208
    88   207
    87   208
    88   208
    88   208
    87   207

How do you want to handle that?

[Rafat]

Thank you so much Voss!!

I want the output in .tif or .nc format to plot it for spatial variability.

Cheers,

raf

[Voss]

You're welcome!

How do you want to handle the situation where the images have different sizes? That is, you say

input: 15 .tif files x m rows x n columns

output: so I want the out put be 1 median x m rows x n columns

but the m and n are not the same for all files. Please advise.

[Rafat]

The differences between the neighbors are not too much (since they are medians and on the same day), so please consider all the same size 87 x 207.

Cheers

[Rafat]

or min m and min n

[Walter Roberson]

If you have 120 files in the directory then how do you want to select the 15 of them you want to process?

[Rafat]

I gave an example here for 15

Answer 1

[Walter Roberson]

format long g
master_size = [87, 207];

unzip Test.zip

Warning: Cannot open file "" for reading.

files = dir('*.tif');
filenames = fullfile({files.folder},{files.name});
N = numel(files);
for ii = 1:N
    files(ii).data = imresize(imread(filenames{ii}), master_size);
end
datablock = cat(3, files.data);

image_median = median(datablock, 3, 'omitnan');
[min(image_median(:)), max(image_median(:))]

ans = 1×2

1.0e+00 * -1.06638e+13 7.329353e+11

image(image_median)

After that you want to write the median out to a tiff, and you probably want the same data range (rather than converting the image to uint8). imwrite() cannot itself write single precision tiff, so you need the Tiff library. More information is at https://www.mathworks.com/matlabcentral/answers/7184-how-can-i-write-32-bit-floating-point-tifs-with-nans . Or you could use the File Exchange contribution https://www.mathworks.com/matlabcentral/fileexchange/30519-export-image-to-tif-or-tiff-file-of-selected-data-type which you could install using the Add-On Explorer

[Rafat]

Dear Walter,

Thank you very much for woderful script!!

I run it, it's doing great!!

one more thing please, the "FillValue = -9.9999998E12" in each .tif file, can we replace it with "NaN" before start finding the median?.

Cheers

[Rafat]

It's my mistake, I thought the FillValue is NaN

[Walter Roberson]

You might be interested in standardizeMissing

format long g
master_size = [87, 207];

unzip Test.zip

Warning: Cannot open file "" for reading.

files = dir('*.tif');
filenames = fullfile({files.folder},{files.name});
N = numel(files);
for ii = 1:N
    thisimg = imread(filenames{ii});
    minval = min(thisimg(:));
    if minval < -1e12
        thisimg = standardizeMissing(thisimg, minval);
    end
    files(ii).data = imresize(thisimg, master_size);
end
datablock = cat(3, files.data);

image_median = median(datablock, 3, 'omitnan');
[min(image_median(:)), max(image_median(:))]

ans = 1×2

45.09927 214.264

image(image_median)

[Rafat]

Hello Walter,

I went through the script, then I understand the wierd values other than the 1- "FillValues" and 2- the normal range of data from 0-1000

"master_size = [87, 207];"

this produce the wierd values in .tif files if rows> 87 and columns > 207; so to that end, I tested

"master_size = [88, 208];" for .tif files if rows= 88 and columns = 208; the wierd values disappierd, so I think we need to deliet the rows > 87 and the columns > 207 then apply from "master_size = [87, 207];" to the end of script.

I appriciat any help of that!!

Cheers,

rafat