Image steganography in matlab:Retrieve hidden Image

I am trying to retrieve the image hidden in another image. This is the code I have so far but it is not working correctly. Can anyone tell me what Im doing wrong?

 I1 = imread('c:/df/img/Lena1.png');
 I2=I1;

 output_stream="";
 data_size_s="";     %In bytes and as a string  

 %We read the size of the target from the stego image (first 16 bits)
 k=1;
for i=1:rows(I2)
   for j=1:columns(I2)
     if (k<18) s=dec2bin(double(I2(i,j,1)),8); data_size_s=strcat(data_size_s,s(8)); k++; endif
     if (k<18) s=dec2bin(double(I2(i,j,2)),8); data_size_s=strcat(data_size_s,s(8)); k++; endif
   if (k<18) s=dec2bin(double(I2(i,j,3)),8); data_size_s=strcat(data_size_s,s(8)); k++; endif
   end
end

 %We pass the size of the target image from string to integer
target_rows=uint32(bin2dec(substr(data_size_s, 1,8)));   
target_columns=uint32(bin2dec(substr(data_size_s,9,8))); 
data_size_int=target_rows*target_columns;  % Data Size in bytes as integer

 %We read the output_stream from the stego image
total_data_size=(8*data_size_int)+16;    % Total length of the output stream (header + target data)

 k=1;
 for i=1:rows(I2)
   for j=1:columns(I2)
     if (k<=(total_data_size)) s=dec2bin(double(I2(i,j,1)),8); output_stream=strcat(output_stream,s(8)); k++; endif
   if (k<=(total_data_size)) s=dec2bin(double(I2(i,j,2)),8); output_stream=strcat(output_stream,s(8)); k++; endif
     if (k<=(total_data_size)) s=dec2bin(double(I2(i,j,3)),8); output_stream=strcat(output_stream,s(8)); k++; endif
   end
end

 I3_rows=uint32(bin2dec(substr(output_stream,1,8)));
 I3_columns=uint32(bin2dec(substr(output_stream,9,8)));
 k=17;

I3=uint8(zeros(I3_rows,I3_columns,3)); 

% We create a empty crey-scale image with the dimensions of the target
total_data_size=(8*data_size_int)+16;

for i=1:I3_rows
  for j=1:I3_columns
      if (k<=total_data_size) I3(i,j,1)=uint8(bin2dec(substr(output_stream,k,8))); k=k+8; endif
     if (k<=total_data_size) I3(i,j,2)=uint8(bin2dec(substr(output_stream,k,8))); k=k+8; endif
      if (k<=total_data_size) I3(i,j,3)=uint8(bin2dec(substr(output_stream,k,8))); k=k+8; endif

    end
  end

 %We show the hidden image.
 imshow(I3);

[Walter Roberson]

How do you know that it is not working properly? Are you getting an error message?

[Jessica]

because I am not getting the correct image as a result. This is the image I am receiving.

[Walter Roberson]

There is nothing obviously wrong with that output. I probably would not have chosen to embed that particular image, but that's just me.

Answer 1

[Image Analyst]

You need to go through the normal debugging process. We don't know your algorithm for hiding and recovering your image, and I don't really have time to learn it, as fascinating as it might be. We can fix an error message for you but we can't do complicated algorithm development.

[Jessica]

I do not have the algorithm used for hiding, It is a class assignment I was given an image and have to recover the hidden image. This is what I am recovering but it is not the correct image.

[Walter Roberson]

How do you know that it is not the correct image?

[Image Analyst]

Did you discuss in class the methods for hiding and recovering? Are you following those processes?

For what it's worth, attached is my demo for LSB watermarking, though it's probably not the method you went over in class.

[Walter Roberson]

For what it's worth, I think you are probably putting the extracted information into the wrong place in the array. When you see exactly three fuzzy repeats of an image you should suspect that you have put three color planes horizontally next to each other instead of stacking them in the third dimension. Consider that if ColorMatrix is something by something by 3 then

reshape( permute(ColorMatrix, [1 3 2]), size(ColorMatrix) )

is going to have the right dimensions to look like a color matrix, but the data will be in the wrong order.

[Image Analyst]

Jessica, can you post both the original "cover" image, and the original "image to be hidden" and which you hope to recover/extract?

Answer 2

[Chris]

I think what you need to do is determine if the image is RGB or Grayscale and then enter a correct loop for processing either a RGB image or Grayscale image. I think the 17th bit in the output_stream will be the determining factor for you.

Answer 3

[riyaz nadaf]

i did this project. if you want matlab code contact

[e-mail]