How to save actual values in a matrix as an image in matlab?

I have a matrix in matlab with 10 different values. All values in my array are integer from 1 to 10. I need to save this matrix as an image (image.jpg). However I want to have the same values ( integers from 1 to 10) in my image when I read the image into the matlab. When I use imwrite or save or saveas I create an image but when I read it into matlab, values are from 0-255 no form 1-10.

Answer 1

[Image Analyst]

imwrite should work. It's not enough that they are integers, they must be uint8 integers, not double values that happen to have integer values. Try this:

imwrite(uint8(myData), 'my image.png');

[marjan ss]

it worked. Thank you so much.

[Image Analyst]

Can you then "Accept this answer"? Thanks in advance.

[Pushkal Sharma]

Will it not change the pixel values to 8 bit (i.e. out of 255)?

[Image Analyst]

Yes, of course. You can use uint16 if you want. If you want it to remain floating point, then I recommend you just use save() to save the variable in a .mat file.

[Pushkal Sharma]

Hi IA! Thanks for getting back to me.

Yeah I think using uint16 solves the issue as it does something similar to ImageJ where it stores 12 bit (i.e. max 4095) matrix in 16 bit format but still assigns the original pixel values i..e not try to rescale them to be in between 0 and 16k (i..e. 16 bit).

Let me know if I understood it wrong.

[Image Analyst]

I always use uint8() and uint16() because I never know if or how the im2uint8() and im2uint16() functions shift intensities or scale the image to fit into the 8 bit or 16 bit range. Like if you have a double with min value of 1000 and max value of 45000, does it leave it as a range [1000-45000] or does it scale it to 0-65535 for uint16? I know that uint8 and uint16 never scale or shift intensity ranges.If they are known to be out of range then I call rescale() or mat2gray() before calling uint8() or uint16(). I prefer to scale it myself so I know what's going on.

[Ayse]

After implementing the suggestions, I am still getting a png fil that is completely black.

My code is:

M=randi([10,93],4,4);

YM=uint8(M);

imwrite(uint8(YM), 'YM.png')

I am trying to save the matrix ouput I see on matlab (like the one below) as a png image as it is.

33 23 77 76

55 91 21 90

90 90 45 65

91 50 86 12

I would appreciate any suggestions,

Thank you.

[DGM]

It depends what the intent is. If the goal is merely to transport the array, then what you've written will work.

M = randi([10,93],4,4); % values are within [0 255] (good!)
YM = uint8(M); % values are unchanged, but class is changed to uint8
imwrite(YM,'YM.png') % write the image

YMrec = imread('YM.png'); % read the image
isequal(YM,YMrec) % arrays are identical

ans = logical
   1

imshow(YM) % display the "image"

I don't see how you'd get a completely black image out of it. The redundant call to uint8() has no effect. If you're not familiar with working with data as images, it's worth understanding that the display (and writing) of image data rests on the assumption of data scale and class. If the data is presented as an integer class, then imshow() and imwrite() will assume that "black" and "white" are represented by the minimum and maximum values which can be represented by the class (e.g. 0-255 for uint8). If the data is presented as a floating-point class (e.g. M in the above example), then black is 0 and white is 1. If you had done

imwrite(M,'M.png')

Then M.png would be an all-white image, and the original values would be unrecoverable.