Image hidding using DWT

Hello every one

I am finding a problime in my code for hidding 325*325 image into 2300*2300 image after decompose it using into HH{3} to get the best PSNR and MSE

and the erroe say that the matrix in not matching.

the code is as follow:-

% Main code
clc
close all;
clear all;
%
a=imgetfile();
coverImage1 = imread(a);
coverImages = imresize(coverImage1,[2600 2600]);
R=coverImages(:,:,1);     %red = 1, green = 2, blue = 3
G=coverImages(:,:,2);
B=coverImages(:,:,3);
coverImage=B;
% figure;
% imshow(coverImage);
% title('Blue Image');
%%%%%
b=imgetfile();
secretImage1 = imread(b);
%secretImage=B1;

secretImage= rgb2gray(secretImage1);
secretImage = imresize(secretImage,[1300 1300]);
row = size(secretImage,1);
col = size(secretImage,2);

s = row*col;
%secretImage = imresize(secretImage,[280 280]);
% figure(1); imshow (coverImages)
% figure(2); imshow (secretImage)

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Perform DNA encryption on the secret image
key = randi([0, 255], size(secretImage)); % Generate random key

encryptedImage = DNAEncrypt(secretImage, key);


%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

nColors = 256;
n =3 ;             %# Number of decompositions
LL = cell(1,n);    %# Approximation coefficient storage
LH = cell(1,n);    %# Horizontal detail coefficient storage
HL = cell(1,n);    %# Vertical detail coefficient storage
HH = cell(1,n);    %# Diagonal detail coefficient storage
 X = coverImage ;
for i = 1:n  %#  %# Apply nLevel decompositions
 [LL{i},LH{i},HL{i},HH{i}] = dwt2(X,'Haar');
 X = LL{i};     % gives you the last quarter of the image 
    
end
  %coverImage = LL{i};     % gives you the last quarter of the image 
  %alpha = 0.02; % Embedding strength
OO= HH{1};  OOO = HH{2}; OOOO=HH{3};% here I want to get the need HH{1,2,3} for the good hidding
%%
 HH_embedded = HH{1} + ( encryptedImage);

stegoImage = idwt2(LL{i}, LH{i}, HL{i}, HH_embedded, 'haar');
imshow(uint8(stegoImage));
% %%
% tiledImage = wcodemat(LL{n},nColors);
% for i = n:-1:1
%   tiledImage = cat(1,cat(2,tiledImage,...
%                            wcodemat(LH{i},nColors)),...
%                      cat(2,wcodemat(HL{i},nColors),...
%                            wcodemat(HH{i},nColors)));
% end
% figure(1);imshow(uint8(tiledImage-1));  
%%
  B2=uint8(stegoImage);


RGB= cat(3, R, G, B2);
stg= double(im2gray(RGB));
% figure (5)
% subplot(1, 2, 1);
% imshow(RGB);
% title('RGB after Analysis');
%%%%
% Display the stego image

figure;
subplot(1, 2, 1);
imshow(coverImages);
title('Cover Image');
subplot(1, 2, 2);
imshow(RGB);
title('Stego Image');
%%%%%%%%%%%%%
%TT = immse(stg, double(coverImage));
mse = sum(sum((coverImage - uint8(stg)) .^ 2)) / numel(coverImage);
% Display the MSE
fprintf('Mean Squared Error (MSE): %.4f\n', mse);
%CC = mse(stgcoverImage);
C  = psnr((uint8(stg)),coverImage);
%fprintf('\n The Mean Sq. Erorr value is %0.4f', CC );
fprintf('\n PSNR value is %0.6f', C);
%%
%

I will write the code here and provide you the images, hope to get it down

[Image Analyst]

I'm not familiar with your algorithm. Does it expect the cover image and secret image to have the same dimensions? Which line throws the error?

By the way, there is a function for mse: immse

[Karbala'a Unvi. Science]

Thank you for your responed.

1- The erorr is in the IDWT where the images and the matrix is not compatable

2- PSNR and MSE are not givinig a good values for the hidding

Pleace tell me if the code is not correct or there is a slution for that

Welling to here from you

Answer 1

[Umar]

% Main code
clc
close all; clear all;
a = imgetfile(); coverImage1 = imread(a); coverImages = imresize(coverImage1, [2600 2600]); R = coverImages(:,:,1);
%red = 1, green = 2, blue = 3
G = coverImages(:,:,2); B = coverImages(:,:,3); coverImage = B;
b = imgetfile(); secretImage1 = imread(b); secretImage = rgb2gray(secretImage1); secretImage = imresize(secretImage, [1300 1300]); row = size(secretImage, 1); col = size(secretImage, 2); s = row * col;
key = randi([0, 255], size(secretImage));
% Generate random key
encryptedImage = DNAEncrypt(secretImage, key);
nColors = 256; n = 3; %# Number of decompositions
LL = cell(1, n); %# Approximation coefficient storage
LH = cell(1, n); %# Horizontal detail coefficient storage
HL = cell(1, n); %# Vertical detail coefficient storage
HH = cell(1, n); %# Diagonal detail coefficient storage
X = coverImage;
for i = 1:n %# Apply nLevel decompositions
    [LL{i}, LH{i}, HL{i}, HH{i}] = dwt2(X, 'Haar'); X = LL{i};
    % gives you the last quarter of the image
end
% Ensure the sizes of LL, LH, HL, and HH match the cover image
LL = cellfun(@(x) imresize(x, size(coverImage)), LL, 'UniformOutput', false); LH = cellfun(@(x) imresize(x, size(coverImage)), LH, 'UniformOutput', false); HL = cellfun(@(x) imresize(x, size(coverImage)), HL, 'UniformOutput', false); HH = cellfun(@(x) imresize(x, size(coverImage)), HH, 'UniformOutput', false);
HH_embedded = HH{1} + encryptedImage; stegoImage = idwt2(LL{i}, LH{i}, HL{i}, HH_embedded, 'haar'); imshow(uint8(stegoImage));
B2 = uint8(stegoImage); RGB = cat(3, R, G, B2); stg = double(im2gray(RGB));
figure; subplot(1, 2, 1); imshow(coverImages); title('Cover Image'); subplot(1, 2, 2); imshow(RGB); title('Stego Image');
mse = sum(sum((coverImage - uint8(stg)) .^ 2)) / numel(coverImage); fprintf('Mean Squared Error (MSE): %.4f\n', mse);
C = psnr(uint8(stg), coverImage); fprintf('\n PSNR value is %0.6f', C);

[Umar]

let's address the issue you mentioned regarding the matrices not matching. The error is most likely occurring in the line where you are performing the inverse discrete wavelet transform (idwt2) to obtain the stego image: stegoImage = idwt2(LL{i}, LH{i}, HL{i}, HH_embedded, 'haar'); The error is likely due to the mismatch in the sizes of the LL, LH, HL, and HH matrices. To resolve this issue, you need to ensure that the sizes of these matrices match the size of the cover image.

To fix this, you can modify your code as follows:

[Umar]

In the modified code, we use the imresize function to resize the LL, LH, HL, and HH matrices to match the size of the cover image. This ensures that the matrices have the same dimensions and resolves the mismatch error.

Please try running the modified code and let me know if you encounter any further issues.

[Karbala'a Unvi. Science]

Dear Sir

It still gives the same error that I had the last time.

If you wish you can try it with any two images of your choice and try both codes and figure out the mistakes that I may have done

The error message is

"Arrays have incompatible sizes for this operation.

Error in untitled (line 40)

HH_embedded = HH{1} + encryptedImage; stegoImage = idwt2(LL{i}, LH{i}, HL{i}, HH_embedded, 'haar');

[Karbala'a Unvi. Science]

Dear Sir

It still gives the same error that I had the last time.

If you wish you can try it with any two images of your choice and try both codes and figure out the mistakes that I may have done

The error message is

"Arrays have incompatible sizes for this operation.

Error in untitled (line 40)

HH_embedded = HH{1} + encryptedImage; stegoImage = idwt2(LL{i}, LH{i}, HL{i}, HH_embedded, 'haar');

[Umar]

My comments: You can check the sizes of the arrays involved in the addition operation to identify the mismatch. Here is an example of how you can check the sizes of the arrays in Matlab: size_HH = size(HH{1}); size_encryptedImage = size(encryptedImage); disp(size_HH); disp(size_encryptedImage); After checking the sizes of HH{1} and encryptedImage, you can adjust the dimensions of one of the arrays to make them compatible for the addition operation. You can use functions like size() to check the dimensions and ensure compatibility before proceeding with the operations.

[Umar]

Here is an example of how you can modify the code snippet to address this issue:

%Check dimensions and resize if needed

if ~isequal(size(HH{1}), size(encryptedImage))

% Resize or reshape arrays to make them compatible

encryptedImage = imresize(encryptedImage, size(HH{1})); end

% Perform element-wise addition after ensuring compatibility HH_embedded = HH{1} + encryptedImage; stegoImage = idwt2(LL{i}, LH{i}, HL{i}, HH_embedded, 'haar');

[DGM]

Will you please stop pasting unreadable wads of unformatted code? Use the code formatting. Use line breaks, or rather, make sure that you don't strip the linebreaks when you paste it from whereever you got the code.

[Umar]

Thank you for bringing this to my attention. I apologize for any frustration caused by the formatting of the code in my previous communications. I will make sure to use proper code formatting, including line breaks, in all future correspondences. Your feedback is valuable to me, and I appreciate your understanding. Please feel free to reach out if you have any further suggestions or concerns. Thank you for your patience.

[Karbala'a Unvi. Science]

Dear Umar,

When I checked the code that you sent, I found out that the last image (X) has the value of 325*325, which is not the same as the encryptedImage that has the size 1300*1300 and that is the mistake that creates the error. I hope that you can fix that and thank you so much for your response for my request

I tried that but know the error is in line 61 when recompine the images to retuern the stigo image to make ready to be compered with the Original image, I will sne the code back to you that you can find out the error reson

% Main code
clc 
close all; clear all;
a = imgetfile(); 
coverImage1 = imread(a);
coverImages = imresize(coverImage1, [2600 2600]);
R = coverImages(:,:,1); %red = 1, green = 2, blue = 3;
G = coverImages(:,:,2); 
B = coverImages(:,:,3);
coverImage = B;
b = imgetfile(); 
secretImage1 = imread(b);
secretImage = rgb2gray(secretImage1);
secretImage = imresize(secretImage, [1300 1300]);
row = size(secretImage, 1);
col = size(secretImage, 2);
s = row * col;
%
key = randi([0, 255], size(secretImage)); % Generate random key 
encryptedImage = DNAEncrypt(secretImage, key);
%%
nColors = 256;
n = 3; %# Number of decompositions 
LL = cell(1, n); %# Approximation coefficient storage 
LH = cell(1, n); %# Horizontal detail coefficient storage 
HL = cell(1, n); %# Vertical detail coefficient storage 
HH = cell(1, n); %# Diagonal detail coefficient storage 
X = coverImage;
for i = 1:n %# Apply nLevel decompositions 
    [LL{i}, LH{i}, HL{i}, HH{i}] = dwt2(X, 'Haar');
    X = LL{i}; % gives you the last quarter of the image 
    OO = HH{3};
 end
% Ensure the sizes of LL, LH, HL, and HH match the cover image 
 LL = cellfun(@(x) imresize(x, size(coverImage)), LL, 'UniformOutput', false);
 LH = cellfun(@(x) imresize(x, size(coverImage)), LH, 'UniformOutput', false);
 HL = cellfun(@(x) imresize(x, size(coverImage)), HL, 'UniformOutput', false);
 HH = cellfun(@(x) imresize(x, size(coverImage)), HH, 'UniformOutput', false);
%
%Check dimensions and resize if needed
if ~isequal(size(HH{1}), size(encryptedImage))
% Resize or reshape arrays to make them compatible
encryptedImage = imresize(encryptedImage, size(HH{1})); 
end

%%
% Perform element-wise addition after ensuring compatibility
HH_embedded = HH{3} + encryptedImage;
stegoImage = idwt2(LL{i}, LH{i}, HL{i}, HH_embedded, 'haar');


%%

% HH_embedded = HH{i} + encryptedImage; 
% stegoImage = idwt2(LL{i}, LH{i}, HL{i}, HH_embedded, 'haar');
% imshow(uint8(stegoImage));


%%
B2 = uint8(stegoImage);
RGB = cat(3, R, G, B2);
stg = double(im2gray(RGB));
figure; subplot(1, 2, 1);
imshow(coverImages); title('Cover Image');
subplot(1, 2, 2);
imshow(RGB); title('Stego Image');
mse = sum(sum((coverImage - uint8(stg)) .^ 2)) / numel(coverImage); fprintf('Mean Squared Error (MSE): %.4f\n', mse);
C = psnr(uint8(stg), coverImage); fprintf('\n PSNR value is %0.6f', C);

[Umar]

Dear friend, To address the bug and enhance the steganography process, the reconstruction of the stego image needs to be corrected. The correct wavelet coefficients should be used to reconstruct the stego image accurately. Below is the corrected code snippet for the stego image reconstruction:

[Umar]

% Corrected stego image reconstruction
HH_embedded = HH_embedded; % Use the correct variable for embedding
stegoImage = idwt2(LL{3}, LH{3}, HL{3}, HH_embedded, 'haar'); % Reconstruct stego image using the correct coefficients

[Umar]

I tried my best, hope this will help you figure out the problem.

[Umar]

the reconstruction of the stego image needs to be corrected. The correct wavelet coefficients should be used to reconstruct the stego image accurately.

[Karbala'a Unvi. Science]

thank you for the help