how can save gray scale image without loss information ?

Hello,

I converted a matrix to a grayscale image.

I saved the image with imwrite

I read the image with imread

The data obtained (uint8) between 0 and 255. in order to convert them to double (0,1) I used im2double.

The result obtained (I3) is a little different to the initial data (I). Is it possible to avoid this error and obtain the same result (I3 = I)?

Thank you

A =[0 200 400 600; 800 1000 1200 1400;1600 1800 2000 2200];
Amin=min(min(A));
Amax=max(max(A));
I=mat2gray(A,[Amin Amax]);
imwrite(I,'NewFile.png')
I2=imread('NewFile.png');
I3= im2double(I2);
isequal(I,I3)

Answer 1

[darova]

Try this

[K. Taieb]

Thank you

It's the same problem

[darova]

THere is a difference (small), indeed

>> A/max(A(:))*255
ans =
         0   23.1818   46.3636   69.5455
   92.7273  115.9091  139.0909  162.2727
  185.4545  208.6364  231.8182  255.0000

I think it's because imread or imwrite rounds the data

  A =[0 200 400 600; 800 1000 1200 1400;1600 1800 2000 2200];
A1 = A/max(A(:));
imwrite(A1,'im.png')
A2 = imread('im.png');
A3 = im2double(A2);

imagesc(A3-A1)
colorbar

[K. Taieb]

Thank you

yes I think too imwrite rounds the data.

Just how I can find the matrix A from the saved image im?

[darova]

Maybe you can't? Hm, i don't know

[K. Taieb]

Thanks :)

Answer 2

[Image Analyst]

I never use im2double(). If it's double, just use it as double. If it's integer and you want double, just use double(). If you want to store as double, use save() and save it to a .mat file. If you want an integer file so you have nice little icons in File Explorer, cast to uint16 or uint8. If the number is beyond those ranges, you can use rescale() or mat2gray() to scale the data before casting to integer.