Help embedding secret message using lwt .cannot extract the secret message correctly

plz help.. here is my code at below

i embedded the secret message in the 1 LSB of each CA

then i did ilwt for CA, CD and converted to uint8

BUT.. when i do lwt again ..then extracting the message from 1LSB of each CA..the extracted message is not the same as the input

BUt it is the same if i did not covert the ilwt to uint8...why?..i should convert it to uint8 because in the advanced code i will assign the result to the row of the image again

% reading the image

img=imread('c:\head4.tif');

% the row (the host row) in which i want to embed the secret message as an example at row
% 305 and start column 3 to end column 570

row_img=img(305,3:570);

% secret message is a vector

secret=[100,200,20,40,50,80,60,230,140,1,2,0,210,205,190,9,0,32,80,100,200,201,202,203,199,189,187,180,170,0,1,0,1,0,100];

secret_bin=dec2bin(secret,8);

secret_bin=secret_bin';

length_secret = size(secret_bin,1)*size(secret_bin,2);

   wname = liftwave('haar','int2int');
   host_row=double(row_img);

   [CA, CD]=lwt(host_row,wname);

   length_CA = size(CA,2);
   c = 1;

   % embedding the secret_bin in the last signficant bit of each CA ( 1 LSB of CA )
   for b = 1:1:length_CA
             if b <= length_secret 
                      xb=CA(1,c);

                       if xb < 0 
                          xb= abs(xb);

                         if strcmp(secret_bin(b), '0')
                            xb=bitset(xb,1,0);
                         else
                            xb=bitset(xb,1,1);

                         end    
                         CA(1,c)=-xb;
                      else
                         if strcmp(secret_bin(b), '0')
                            CA(1,c) = bitset(CA(1,c),1,0);
                         else
                            CA(1,c) = bitset(CA(1,c),1,1);

                         end    
                      end% end if xb<0

                      c = c+ 1;
                       if (c > length_CA)
                         c = 1;
                       end
              end  % if b <= length_secret 
   end % end for loop b=8:-1:1

                    stego_host_uint8 = uint8(ilwt(CA,CD,wname));
                    stego_host_dble = (ilwt(CA,CD,wname));

   % Extracting the secret vector from the stego_host_uint8   
    watermarked_host=double(stego_host_uint8);

    [CA_ex, CD_ex] = lwt(watermarked_host,wname);  

    length_CA_ex = size(CA_ex,2);

    c = 1;

    secret_bits_ex = zeros(1,length_secret);

      for b = 1:1:length_CA_ex

          if b <= length_secret
               xb=CA_ex(1,c);

               if xb < 0 
                 xb= abs(xb);
               end
             secret_bits_ex(b) = bitget(xb,1);
          %letter = bitset(letter,b,bitget(xb,1));
             c = c + 1;
          end
      end

      % function i created
      secret_ex=binRoundC2Dec(secret_bits_ex,8);

      % comparison between the input secret vector and the extracted
      message=secret';

      if secret_ex == message
          disp('yes equal');
      else
          disp('No Equal');
      end

      % the extracted output [100;200;20;40;114;80;60;230;140;1;2;0;210;205;190;9;0;32;80;100;200;201;202;203;199;189;187;180;162;0;1;0;1;0;100]
      % the input secret     [100;200;20;40;50;80;60;230;140;1;2;0;210;205;190;9;0;32;80;100;200;201;202;203;199;189;187;180;170;0;1;0;1;0;100]
      % they are diffrent at positions 5 , 29 ...why?

      % Extracting the secret vector from the stego_host_dble   
    watermarked_host_dbl=double(stego_host_dble);

    [CA_ex_dbl, CD_ex_dbl] = lwt(watermarked_host_dbl,wname);  

    length_CA_ex = size(CA_ex_dbl,2);

    c = 1;

    secret_bits_ex_dbl = zeros(1,length_secret);

      for b = 1:1:length_CA_ex

          if b <= length_secret
               xb=CA_ex_dbl(1,c);

               if xb < 0 
                 xb= abs(xb);
               end
             secret_bits_ex_dbl(b) = bitget(xb,1);
          %letter = bitset(letter,b,bitget(xb,1));
             c = c + 1;
          end
      end

      % function i created
      secret_ex_dbl=binRoundC2Dec(secret_bits_ex_dbl,8);

      % comparison between the input secret vector and the extracted 
      message=secret';

      if secret_ex_dbl == message
          disp('yes equal from stego dble');
      else
          disp('No Equal from stego dble');
      end

      % the extracted output [100;200;20;40;50;80;60;230;140;1;2;0;210;205;190;9;0;32;80;100;200;201;202;203;199;189;187;180;170;0;1;0;1;0;100]
      % the input secret     [100;200;20;40;50;80;60;230;140;1;2;0;210;205;190;9;0;32;80;100;200;201;202;203;199;189;187;180;170;0;1;0;1;0;100]
      % they are equal when extracting directly from double lwt without  conversion to uint8

[Walter Roberson]

You do not appear to have posted binRoundC2Dec

[shimaa samy]

it is a simple function

function [ roundc_dec] = binRoundC2Dec( roundc_binary,base )

    size_bin_roundc=size(roundc_binary,2);

    if mod(size_bin_roundc,base)==0
        size_roundc= size_bin_roundc /base;
        roundc_dec=zeros(size_roundc,1);

        start=1;
        for i=1:1:size_roundc

            roundc_dec(i,1)=bin2dec(num2str(roundc_binary(start:base*i)));
            start=base*i+1;

        end    

    else
        disp('the size bits of data can not be divided on ');
        disp(base);
        disp(' and size data bits = ')
        disp(size_bin_roundc);
    end

end

[shimaa samy]

plz help me :) ... till NOW i can not retrieve the secret message correctly and i do not know which line is wrong in my code

[Walter Roberson]

Sorry, my substitute sleepers quit on me and I have to do all my sleeping myself these days.

[shimaa samy]

i did not understand your expression BUT thanx

[Walter Roberson]

I do not have your image, so I used the demonstration image AT3_1m4_02.tif instead. The output I get is

yes equal
yes equal from stego dble

[shimaa samy]

Here is my cover image, but how it works with u, it is working with me with the stego_host_dble ( it is for this message ' yes equal from stego dble' ) but not working when using stego_host_uint8

my cover image is also uint8

[shimaa samy]

i attached my code files... i appreciate your reply ...but it is still NOT working with me

[rizky alvarez]

excuseme shimaa, do you find the solution for this problem?

i have same problem with the code

Answer 1

[Walter Roberson]

With that head4.tif file, there are three places where stego_host_uint8 is different from stego_host_dble . In two of the places, stego_host_dble is -1, which is outside of the range of uint8 and would be converted to 0 when you used uint8(). The other different location is 256, which is also outside of the range of uint8 and would be converted to 255 when you used uint8()

Your code contains a mistaken assumption: the assumption that if you take the wavelet coefficients of an image in the range 0 to 255, and you alter the wavelet coefficients by no more than +/- 1 in each position, that when you take the inverse wavelet transform with the resulting coefficients that the result will be strictly in the range 0 to 255.

Another way of phrasing this is you assume that every original matrix host_row that has a value outside of the range 0 to 255 will definitely result in a CA that has a value outside of that range -- because if there is an original matrix host_row that has values outside of that range that transforms into a CA that has values strictly inside the range, then the implication is that a modified CA transformed through inverse wavelet can end up having values outside of the desired range.

My tests show that is not the case: that there are simple matrices with values outside 0 to 255 that transform into a CA that has values strictly inside the range.

Simple example:

home_row = zeros(1, 568); home_row(1) = 255;
[CA, CD] = lwt(home_row, wname);
CA(1)     %will be 128
CA(1) = CA(1)+1;   %make it 129
back = iwt(CA, CD, wname);
back(1)   %will be 256

So, although it is sometimes possible to alter the coefficients by +/- 1 and have the result of the ilwt be strictly 0 to 255, it is not always possible to do so. I do not know what the conditions are under which it can work.

[shimaa samy]

your analysis is right..thank u alot :) should i check values or make any condition before embedding in the 1LSB CA to solve this problem ?. Or, there is another way for embedding vector with range [0 to 255] inside row of image with range [0-255]?

i am very thankful for your helping

[Walter Roberson]

I do not know what you should do.

I believe you should be able to use iwt() (Integer Wavelet Transform)

[Ashmiya Stephen]

I`m beginner in matlab. I used the same code for extraction and embedding process and it resulted in error. please hep me.. the error code is as follows

img=imread('frame1.jpg'); 
row_img=img(305,3:570); 
secret='h'; 
secret_bin=dec2bin(secret); 
disp(secret_bin); 
length_secret = size(secret_bin,1)*size(secret_bin,2); 
wname = liftwave('haar','int2int'); 
host_row=double(row_img); 
[CA, CD]=lwt(host_row,wname); 
length_CA = size(CA,2); c = 1; 
% embedding the secret_bin in the last signficant bit of each CA ( 1 LSB of CA ) 
for b = 1:1:length_CA if b <= length_secret  xb=CA(1,c); 
if xb < 0  xb= abs(xb); if strcmp(secret_bin(b), '0') xb=bitset(xb,1,0); else xb=bitset(xb,1,1); 
end     
CA(1,c)=-xb; 
else if strcmp(secret_bin(b), '0') CA(1,c) = bitset(CA(1,c),1,0); 
else CA(1,c) = bitset(CA(1,c),1,1); 
end     
end% end if xb<0 c = c+ 1; 
if (c > length_CA) c = 1; 
end end  % if b <= length_secret 
 end % end for loop b=8:-1:1 stego_host_uint8 = uint8(ilwt(CA,CD,wname)); stego_host_dble = (ilwt(CA,CD,wname));
 % Extracting the secret vector from the stego_host_uint8    watermarked_host=double(stego_host_uint8); 
[CA_ex, CD_ex] = lwt(watermarked_host,wname);   
length_CA_ex = size(CA_ex,2); 
c = 1; secret_bits_ex = zeros(1,length_secret); 
for b = 1:1:length_CA_ex if b <= length_secret xb=CA_ex(1,c); 
if xb < 0  xb= abs(xb); 
end secret_bits_ex(b) = bitget(xb,1); %letter = bitset(letter,b,bitget(xb,1)); 
c = c + 1; end end % function i created result=secret_bits_ex; disp("extracted binary") disp(result); 
a = char(bin2dec(result)) 
disp(a); 

[Walter Roberson]

As we do not have yur input image to test with, you will need to tell us exactly what the error message is.

Answer 2

[PRATAP MANDAL]

I have tried to execute the code, but the code is not working. It is saying "undefined function or variable 'binRoundC2Dec' ". Can anybody help me in this regard ?

[Walter Roberson]

It was posted in https://www.mathworks.com/matlabcentral/answers/338974-help-embedding-secret-message-using-lwt-cannot-extract-the-secret-message-correctly#comment_451639

You may need to Expand Comments