2

I am reading the documentation of the RGB2IND function, and found that it uses Uniform Quantization, Minimum Variance Quantization, and Inverse Colormap, though I don't quite get how these fit into the algorithm given the parameters.

Are there any code examples in any interpreted language (MATLAB, JavaScript, Python) that demonstrate exactly how this function works in this syntax?

[X,map] = rgb2ind(RGB,n);
Amro
  • 123,847
  • 25
  • 243
  • 454
Patrick Roberts
  • 49,224
  • 10
  • 102
  • 153

1 Answers1

6

As the documentation page says, it depends on which form of the function you use:

  • [X,map] = rgb2ind(RGB,n): if you specify the number of colors as input, this will use minimum variance quantization to build an indexed image with at most n colors

  • [X,map] = rgb2ind(RGB,tol): if you specify a tolerance value as input, it uses uniform quantization to build indexed image with at most (floor(1/tol)+1)^3 colors

  • X = rgb2ind(RGB,map): if you specify a colormap as input, it will use the inverse colormap algorithm to build indexed image mapped to the specified map

You could always read the source code yourself (edit rgb2ind)

Here are examples showing how to use all forms of the functions:

%% some truecolor image
RGB = imread('pears.png');
imshow(RGB)

%% 16 colors
[X,map] = rgb2ind(RGB, 16);
imshow(X,map)

%% 0.15 tolerance, no dithering
[X,map] = rgb2ind(RGB, 0.15, 'nodither');
imshow(X,map)

%% use a pinkish colormap with 32 colors
map = pink(32);
X = rgb2ind(RGB, map);
imshow(X,map)
Amro
  • 123,847
  • 25
  • 243
  • 454
  • This is just the documentation... I think the OP is asking for a reference or explanation of these techniques. – Alec Jacobson Mar 30 '15 at 22:34
  • See this: https://www.mathworks.com/help/images/reducing-the-number-of-colors-in-an-image.html. Also related functions like: `imapprox` and [`dither`](https://en.wikipedia.org/wiki/Floyd%E2%80%93Steinberg_dithering) – Amro Oct 29 '16 at 21:00