Convert RGB image into indexed image and save it

Question

I have an RGB image. So when I do [image2, map] = imread('image.png') in Matlab I get [] for map as expected. I want to convert that RGB image to an indexed image. I want to save it as an indexed image with one channel and see colours like here.

Referring here I used the following code in Matlab.

image2 = imread('image.png');
IND = rgb2ind(image2,256);
imwrite(IND, 'imageIndexed.png')

But what is saved is a grey-scale image. And when I read it back still the map is []. I want to write it so that it will be a colour image and when I use [image2, map] = imread('image.png') next time, map won't be []. Can someone please help?

Answer 1

You can do the following:

• Convert image from RGB to indexed image with 2 color:

  [X, cmap] = rgb2ind(RGB, 2);
  

• Replace indices of color map to black and white:

  cmap(1, 1:3) = [0, 0, 0]; %Fist color is black
  cmap(2, 1:3) = [1, 1, 1]; %Second color is white
  

• Write indexed image (and map) to PNG file:

  imwrite(X, cmap, 'K.png');
  

Pay attention: you need to write the matrix and the color map, when writing and indexed image to file.

• Read image (and map) from PNG file, and convert it to RGB image:

  [I, cmap2] = imread('K.png');
  L = ind2rgb(I, cmap2);
  

---

Here is a code sample:

RGB = imresize(imread('peppers.png'), 0.5);   %Read input RGB image.

%Convert image to indexed image with 2 color.
[X, cmap] = rgb2ind(RGB, 2);
J = ind2rgb(X, cmap);

%Replace indices of color map:
cmap(1, 1:3) = [0, 0, 0]; %Fist color is black
cmap(2, 1:3) = [1, 1, 1]; %Second color is white

K = ind2rgb(X, cmap);

figure;imshow(RGB);
figure;imshow(J);
figure;imshow(K);

imwrite(X, cmap, 'K.png');

[I, cmap2] = imread('K.png');
L = ind2rgb(I, cmap2);
figure;imshow(L);

---

For completion, here is an example using your reference:

[webX, web_cmap] = imread('https://i.sstatic.net/zuIra.png');   %Read input image and color map.
RGB = ind2rgb(webX, web_cmap);

%Convert image to indexed image with 4 colors.
[X, cmap] = rgb2ind(RGB, 4);

%Collect histogram
H = histogram(X, 4);
H = H.Values';

%Replace indices of color map: set the color with minimal pixels to white, and other to black.
cmap(H == min(H), :) = 1; %Fist color is black
cmap(H ~= min(H), :) = 0; %Set other three colors to black

%Convert to RGB
bwRGB = ind2rgb(X, cmap);

%Convert to indexed image with only 2 colors:
[X, cmap] = rgb2ind(bwRGB, 2);

imwrite(X, cmap, 'K.png');

[I, cmap2] = imread('K.png');
L = ind2rgb(I, cmap2);
figure;imshow(L);

Result: