imwrite and imread are not giving the same matrix

Hello,

I want to show a matrix in an image (and save it for later use) and then retrieve the matrix from that image using imread. I followed the approach presented in https://www.mathworks.com/matlabcentral/answers/242747-how-can-i-convert-a-matrix-with-decimal-values-into-a-image-and-then-retrieve-the-same-matrix-back-f

However, while saving the image with the command imwrite, and then later read it with imread, I see two different matrices. Here is what I have done:

clear all, close all, clc

% A=rand(10); %random matrix(10x10) with values between 0 to 1.

% B=A*150;

B = [3 1; 2 5];

sz = size(B);

R = typecast(uint64(B(:) * 2^56),'uint8');

R1 = reshape(R, [8 sz]);

R2 = permute(flipud(R1), [2 3 1 4]);

R3 = reshape(R2, [sz(1), sz(2)*8, 1]);

imshow(R3);

imwrite(R3, 'TestFile.jpg')

close all

A = imread('TestFile.jpg');

sz = size(A);

U3 = reshape(A, [sz(1), sz(2)/8, 8, 1]);

U2 = flipdim(permute(U3, [3 1 2 4]),1);

UB = reshape(double(typecast(U2(:), 'uint64')) / 2^56, [sz(1), sz(2)/8, 1]);

Now my question is how to get the correct matrix from imread.

Thank you.

Khalid

Answer 1

[Walter Roberson]

This is normal and expected for image formats that use lossy compression, such as the way most people use jpeg.

You can:

• tell imwrite to use jpeg lossless mode; or
• use an image file format such as png that is lossless; or
• arrange the data ahead of time to be flawed in exactly the same way that jpeg throws away data so that what gets written out matches what gets read in; or
• adjust your expectations so that you no longer expect them to match.

I recommend that you adjust your expectations, and because of that, you stop using jpg for scientific work (jpeg2000 is a different matter and worth considering.)

[Khalid Ibne Masood]

Thank you, it worked perfectly.

Thanks again.

Khalid