The Still Image compression For Effective Use Of Bandwidth

Sir,

My project is "The Still Image compression For Effective Use Of Bandwidth". I have sucessfully compiled my code for gray scale images but while trying to compress color images getting an error that to while calculating mean square error(MSE) and peak signal to noise ratio (PSNR)indicating " ??? Error using ==> minus Number of array dimensions must match for binary array op. So, please kindly reply me soon what is the problem in the code and why it is possible to calculate MSE and PSNR only for gray scale images and not for color images.

MY code is

%%reading an image
input_image=imread('1.png');  
input_image = imresize(input_image,[256 256]);
 imshow(input_image);
handles.input_image=input_image;
[r c p]=size(input_image);

        if p==3   %%%if p=3 then it is coler image 
            warndlg('convert color image into Gray image');
            input_image=rgb2gray(input_image);  %%%converting color image into gray image 
        end

      input_image=imresize(input_image,[256 256]);

      input_image=double(input_image);
   [LL LH HL HH]= dwt2(input_image,'haar');
   new_image=[...
                LL  LH
                HL  HH ...
              ];

         figure, imshow(new_image,[]);
                 handles.new_image=new_image;
      image_LL=handles.new_image;

[row1 col1] = size(image_LL);

Image_LL_resize = round(reshape(image_LL,[1 row1*col1]));
 New_Image_LL_resize= round(Image_LL_resize/10);

   %%%%%%%%% Huffman Encoding  %%%%%%%%%%%

   leng = length( New_Image_LL_resize);
  maxr = max( New_Image_LL_resize);
  minr = min( New_Image_LL_resize);
  j = 1;
  for i = minr:maxr
      N = find(New_Image_LL_resize==i);
      count = length(N);
      probab(j) = count/leng;
      j = j+1;
  end
   size_of_dict = length(probab);
  [dictval avglen] = huffmandict([minr:maxr],[probab]);
  y = huffmanenco( New_Image_LL_resize,dictval);

bitstream_image = y;

%%%%% finding length of bitstream  %%%%%
bitstream = length(y);
disp('Length of Compressed Huffman Bitstream');
display(bitstream);

handles.dictval=dictval;
handles.y=y;
warndlg('Encoding Process Completed');

%%decoding
dictval=handles.dictval;
y=handles.y;

Dec_Out = huffmandeco(y,dictval);
len=length(Dec_Out);
row1=sqrt(len);
col1=row1;
New_Decim = reshape(Dec_Out,[row1 col1]);
New_Decim = New_Decim*10;
% figure(1),imshow(New_Decim,[]);
warndlg('Decoding Proces Completed');
handles.New_Decim = New_Decim;

%%Inverse DWT
Rec_image = handles.New_Decim;
%apply idwt

% img_idwt = idwt2(
[rows cols] = size(Rec_image);
for i = 1:rows/2
    for j = 1:cols/2
        CA(i,j) = Rec_image(i,j);
    end
end
for i = rows/2+1:rows
    for j = 1:cols/2
        CH(i-rows/2,j) = Rec_image(i,j);
    end
end
for i = 1:rows/2
    for j = cols/2+1:cols
        CV(i,j-cols/2) = Rec_image(i,j);
    end
end
for i = rows/2+1:rows
    for j = cols/2+1:cols
        CD(i-rows/2,j-cols/2) = Rec_image(i,j);
    end
end
% figure,imshow(CA,[]);
% figure,imshow(CH,[]);
% figure,imshow(CV,[]);
% figure,imshow(CD,[]);
output_image = idwt2(CA,CH,CV,CD,'haar');
%axes(handles.axes4);
figure, imshow(output_image,[]);
handles.output_image = output_image;

%%MSE
img1 = handles.input_image;
img2 = handles.output_image;
img2=imresize(img2,[256 256]);
[M N] = size(img1);
img2 = uint8(img2);
MSE = sum(sum((img1-img2).^2))/(M*N);  
set(handles.edit1,'string',MSE);
handles.MSE = MSE;

%%PSNR
PSNR =abs(10*log10(255*255/MSE));

[sailaja]

ur code is so good i am getting output thank u so much

[sailaja]

how to use this code for dct instead of dwt plz help me

Answer 1

[Walter Roberson]

To handle a color image, compress the R, G, and B planes separately.