How do I convert a PNG to a matrix?

I need assistance in converting a PNG file to a matrix. At first, I save downloaded file from a matrix to a PNG file, using the code below:

if true
  % code
fig=figure('visible','off');
imagesc(outo);
colormap(jet(512))
% colormap(jet(256))
x=gca;
x.YDir='normal';
set(x,'position',[0 0 1 1],'units','normalized')
truesize;
caxis([0 55])
axis off
set(fig,'PaperPositionMode','auto')
print('outo','-dpng','-r0')
end

Now, once I have this file called outo.png, I want to return it back to a matrix I originally started with. Down below is the code I use to read that file, my problem is that I do not get the same numbers.

if true
  % code
imgpath2= tt1(i).name;
  map=colormap(jet(1024));
  RGB=imread(imgpath2);
  S=rgb2ind(RGB,map);
  S=double(S);
  L=size(map,1);
  outraw=interp1(1:L,linspace(min_org,max_org,L),S(:,:));
  % pixels around campus
  outraw=flipud(outraw);
end
Does anyone know how to do a better job?  I use jet(1024) instead of jet(256), because it give me better numbers.
I have attached the file I generated and a MAT file with the original matrix, so you can see what the values should be in this 348X348 matrix. I would really appreciate someone coming up with this, I am sure there are many who could use the alsorithm.
Thank you,
Larry

[Suhan]

Hi Larry, I am having troubles in reproducing the issue. You have attached only an array refl_new. Is this the variable 'outo' in the above code snippet?

Also, the code snippet is not very clear. Could you modify the code snippet and also elaborate your query so that all of us can understand it better? For example, what is 'imgpath2' in the code snippet?

[Rik]

It sounds to me as if you want to index an image, so why not using ind2rgb? Writing a file to read it immediately after that is usually a sign you should rethink your approach.

Also, next time, please follow the question guidelines here and here. It will greatly improve your chances of getting an answer.

Answer 1

[Image Analyst]

Onto.png never was a matrix so you can't return it to one. You're saving the whole figure not the image matrix itself. That includes axes, tick mark, etc. If you want to save the image itself and recall it then use imwrite() instead of print().

[Image Analyst]

Larry's reply moved here:

Thank you for looking at this problem. Let me restate the problem, I have a matrix (348 X 348) with double values from 0 to 55. I saved the matrix as a PNG. Now I would like to take the PNG and return it to a matrix for data analysis. I put a MAT file with an image (348 X348) in double form, and I have the PNG of that same image. I provided these two items so one could test the routine they came up with. Take the PNG and bring it back to a matrix. Incidentally, there is no axis information saved, it is a plain image.

To describe my first snippit: The outo (matrix is going to be saved as PND. Unfortunately, I called the PNG file outo.png. Now that saves the file, the other commands prevent printing the image (since I do about 20 of these in the routine, not in this routine. I remove white spacing, axial information, etc. Now the second snippit is used to read the PNG file. Unfortunately, this is the way I address all of the PNG files in a folder. Let us forget about mulple PNG files and work on the one we saved outo.png. I want to bring it back to the original matrix and snippit #2 is my procedure. Impath2 is the image path, since it is not in this directory. I save the map I used to initially create the file except with more resolution, 1024, and I am using the jet palette. Since the PNG file is a RGB file, I convert it to an indexed file then interpolate each pixel. The result is close (0.5-1.3), but I need better accuracy. Can anyone help? Thanks, Larry

[Image Analyst]

Larry, again, you're saving the figure, not the image. The image is saved into the CData property of an "axes" which is in a "figure". Using print saves the figure, not the image inside the axes. Just look at the help: "print(filename,formattype) saves the current figure to a file".

If you want to save and recall a double variable - only in MATLAB and not with any other program - then save the outo array to a proprietary format .mat file with save().

save('outo.mat', 'outo');

If you want to save outo - the actual image, NOT the figure - as an image that any program can read, you must save it in a file format like PNG. So then cast it to integer and save it with imwrite():

uint8Image = uint8(255 * mat2gray(outo));
imwrite(uint8Image, 'outo.png');

[Image Analyst]

Again, you're saving the figure, not the image. The image is saved into the CData property of an "axes" which is in a "figure". Using print saves the figure, not the image inside the axes. Just look at the help: "print(filename,formattype) saves the current figure to a file".

If you want to save and recall a double variable - only in MATLAB and not with any other program - then save the outo array to a proprietary format .mat file with save().

save('outo.mat', 'outo');

To recall:

s = load('outo.mat');
outo = s.outo;

If you want to save outo - the actual image, NOT the figure - as an image that any program can read, you must save it in a file format like PNG. So then cast it to integer and save it with imwrite():

uint8Image = uint8(255 * mat2gray(outo));
imwrite(uint8Image, 'outo.png');

To recall you use imread()

grayImage = imread('outo.png');

but to get the values in the correct range, you're going to have to save the min and max values of it in a .mat file so you can recall them later.

minValue = min(outo(:));
maxValue = max(outo(:));
save('outo.mat', 'minValue', 'maxValue');

To recall:

s = load('outo.mat');
minValue = s.minValue;
maxValue = s.maxValue;
outo = (maxValue - minValue) * double(grayImage / 255) + minValue;

But if you're going to do that, you might as well save outo in the mat file also.