Convert rgb image to an indexed image without decreasing the colors in Matlab

Question

I am doing steganography in indexed image. One of my problems is to convert the rgb image to an indexed image without decreasing the colors. I found out rgb2ind function in matlab converts rgb to indexed using minimum variance quantization and dithering. What I want is to make this indexed image colors exactly like rgb colors. How can i do that?

For the example, in this code i have this matrix M and convert it with rgb2ind function. I'm expecting the result variable Mrgb is equal with the matrix M.

% Matrix M 
M(:,:,1) =  [225  230  250  243   20    3  244  247  255  255
242  252  239    0  239  224   27  252  255  255
224  239   28  243  236  231  240   11  255  255
243  252   15  224   15   12  251    0  255  255
224   11  252  227    0   11  232  251    0  255
243    4  255  228   19    4  243  224    0  255
 0  255  236  251  228  251  224  243  255    0
 19  236  255  224   27   12  251  236  255    0
  0  255  255  255  255  255  255  255  255    0
255    0    0    0    0    0    0    0    0  255];

M(:,:,2) =[255  255  255  255    0    0  255  255  255  255
255  255  255    0  255  255    0  255  255  255
255  255    0  255  255  255  255    0  255  255
255  255    0  255    0    0  255    0  255  255
255    0  255  255    0    0  255  255    0  255
255    0  255  255    0    0  255  255    0  255
0  255  255  255  255  255  255  255  255    0
0  255  255  255    0    0  255  255  255    0
0  255  255  255  255  255  255  255  255    0
255    0    0    0    0    0    0    0    0  255];

M(:,:,3) = [255  255  255  255    0    0  255  255  255  255
255  255  255    0    0    0    0  255  255  255
255  255    0    0    0    0    0    0  255  255
255  255    0    0    0    0    0    0  255  255
255    0    0    0    0    0    0    0    0  255
255    0    0    0    0    0    0    0    0  255
0    0    0    0    0    0    0    0    0    0
0    0    0    0    0    0    0    0    0    0
0    0    0    0    0    0    0    0    0    0
255    0    0    0    0    0    0    0    0  255];

[Mind map] = rgb2ind(M,255);
Mrgb = ind2rgb(Mind, map);
Mrgb=im2uint8(Mrgb);
isequal(M, Mrgb)

UPDATE

I update the matrix M in 4-D. I tried code from Dan's answer and show the image with imshow, but the image become purple when it supposed to be yellow. How can I fix it?

M(:,:,1,1) = [ 225  230  250  243   20    3  244  247  255  255
  242  252  239    0  239  224   27  252  255  255
  224  239   28  243  236  231  240   11  255  255
  243  252   15  224   15   12  251    0  255  255
  224   11  252  227    0   11  232  251    0  255
  243    4  255  228   19    4  243  224    0  255
    0  255  236  251  228  251  224  243  255    0
   19  236  255  224   27   12  251  236  255    0
    0  255  255  255  255  255  255  255  255    0
  255    0    0    0    0    0    0    0    0  255];


M(:,:,2,1) = [  255  255  255  255    0    0  255  255  255  255
  255  255  255    0  255  255    0  255  255  255
  255  255    0  255  255  255  255    0  255  255
  255  255    0  255    0    0  255    0  255  255
  255    0  255  255    0    0  255  255    0  255
  255    0  255  255    0    0  255  255    0  255
    0  255  255  255  255  255  255  255  255    0
    0  255  255  255    0    0  255  255  255    0
    0  255  255  255  255  255  255  255  255    0
  255    0    0    0    0    0    0    0    0  255];


M(:,:,3,1) = [ 255  255  255  255    0    0  255  255  255  255
  255  255  255    0    0    0    0  255  255  255
  255  255    0    0    0    0    0    0  255  255
  255  255    0    0    0    0    0    0  255  255
  255    0    0    0    0    0    0    0    0  255
  255    0    0    0    0    0    0    0    0  255
    0    0    0    0    0    0    0    0    0    0
    0    0    0    0    0    0    0    0    0    0
    0    0    0    0    0    0    0    0    0    0
  255    0    0    0    0    0    0    0    0  255];


M(:,:,1,2) = [  255  255  255  255    0    0  255  255  255  255
  255  255  255    0  255  255    0  255  255  255
  255  255    0  255  255  255  255    0  255  255
  255  255    0  255  255  255  255    0  255  255
  255    0  255  255  255  255  255  255    0  255
  255    0  255  255  255  255  255  255    0  255
    0  255  255  255  255  255  255  255  255    0
    0  255  255  255  255  255  255  255  255    0
    0  255  255  255  255  255  255  255  255    0
  255    0    0    0    0    0    0    0    0  255];


M(:,:,2,2) = [  255  255  255  255    0    0  255  255  255  255
  255  255  255    0  255  255    0  255  255  255
  255  255    0  255  255  255  255    0  255  255
  255  255    0  255  255  255  255    0  255  255
  255    0  255  255  255  255  255  255    0  255
  255    0  255  255  255  255  255  255    0  255
    0  255  255  255  255  255  255  255  255    0
    0  255  255  255  255  255  255  255  255    0
    0  255  255  255  255  255  255  255  255    0
  255    0    0    0    0    0    0    0    0  255];


M(:,:,3,2) = [  255  255  255  255    0    0  255  255  255  255
  255  255  255    0    0    0    0  255  255  255
  255  255    0    0    0    0    0    0  255  255
  255  255    0    0    0    0    0    0  255  255
  255    0    0    0    0    0    0    0    0  255
  255    0    0    0    0    0    0    0    0  255
    0    0    0    0    0    0    0    0    0    0
    0    0    0    0    0    0    0    0    0    0
    0    0    0    0    0    0    0    0    0    0
  255    0    0    0    0    0    0    0    0  255];



[map, ~, Mind] = unique(reshape(M,[],3), 'rows');
[row, col, ~, frame]=size(M);
Mind = reshape(Mind, [row, col, frame]);

%%convert to rgb 
Mrgb=zeros(row, col, 3, frame);
z = size(map,1);
Mrgb = cat(3, map(Mind), map(Mind + z), map(Mind + 2*z));
Mrgb = reshape(Mrgb, [row, col,3, frame]);

map=uint8(map);
map=im2double(map);

imshow(Mind(:,:,1), map)

score 3 · Accepted Answer · edited May 23 '17 at 12:01

3

If you really need a colour resolution of 256^3 then you can do it manually fairly easily, however, you will have to convert your image back to RGB to view it so I'm not sure this actually has any utility:

[map, ~, Mind] = unique(reshape(M,[],3), 'rows');
Mind = reshape(Mind, size(M(:,:,1)));

You can reconstruct your colour image either using sub2ind or like this:

z = size(map,1);
Mrgb = cat(3, map(Mind), map(Mind + z), map(Mind + 2*z))

I'm not sure if this has any point but maybe it's what you want.

isequal(M, Mrgb)

returns 1.

edited May 23 '17 at 12:01

Community

1
1

answered Mar 23 '16 at 08:39

Dan

45,079
17
88
157

1

Once you've got `map`, can't you pass that in to `rgb2ind(M, map)`? (Assuming `numel(map) <= 65536`) – beaker Mar 23 '16 at 16:05
@Dan thanks a lot Dan, yes, this is what i need. Could you help me converting to indexed image with 4-D array, please? I update the matrix in my qusetion. – Fitri Mar 24 '16 at 06:18
@Fitri that's a different question and I suggest you ask it as a separate question. But I would first wonder why you can't just use a simple `for` loop? – Dan Mar 24 '16 at 06:28
@Dan Ok, i create new question for that. I already got answer from my first question. Thanks :) – Fitri Mar 24 '16 at 07:17

Ander Biguri · Answer 2 · 2016-03-23T09:01:14.860

2

EDIT It seems that still, rgb2ind discretizes the colors to a lower number. The valuers are very close (max 4) but still not the same. Another approach (such as @Dan's) is needed.

The second argument of rgb2ind is the amount of colors, with a possible maximum of 65,536. Just change your code to

[Mind map] = rgb2ind(uint8(M),65536,'nodither');

You possibly wont have that many colors.

NOTE Your M is a double in your code, thus rgb2ind doesn't really understand colors above the value 1. Cast it to uint8 and you'll get a map.

edited Mar 23 '16 at 09:01

answered Mar 23 '16 at 08:26

Ander Biguri

35,140
11
74
120

It doesn't make difference in my case. I need the data in variable M and Mrgb after converting is exactly same. – Fitri Mar 23 '16 at 08:43

Convert rgb image to an indexed image without decreasing the colors in Matlab

2 Answers2