How do I draw a colour bar with known discrete RGB values and known percentage of pixels represent each RGB values?

Hi,

I extracted 3 key colours from an image using clustering method. The RGB values of the 3 colours are as below:

RGB1=[24 70 187]; (blue)

RGB2=[208 118 22]; (orange)

RGB3=[88 90 28]; (olive green)

My question is:

How to draw a colour bar (shown below) in matlab with the 3 known colours and the known percentage of the pixels represent each colour (blue 50%, orange 30%, and olive green 20%). In this fashion it tells that the dominant colour extracted from the image is blue followed by orange, then olive green.

Currently this colour bar was drawn in photoshop unfortunately.

Answer 1

[Jan]

As pixel image:

Map = [24 70 187;  208 118 22;  88 90 28] / 255;
Ind = repmat(repelem([1, 2, 3], [50, 30, 20]), 40, 1);
RGB = ind2rgb(Ind, Map);
figure;
image(RGB)

Or as patch:

Map = [24 70 187;  208 118 22;  88 90 28] / 255;
a = 0.5; b = 0.8;
vert = [0 0; 0 1; a 1; a 0; b 1; b 0; 1, 1; 1, 0];
face = [1, 2, 3, 4; 4, 3, 5, 6; 6, 5, 7, 8];
figure;
patch('Vertices', vert, 'Faces', face, ...
      'FaceVertexCData', Map, ...
      'FaceColor', 'flat', 'EdgeColor', 'none')

[Salad Box]

Hi Jan,

Thanks for your answer. It did solve 99.9% of my problem. The only 0.01% unsolved is that with your code for pixel image I get:

Instead of getting the output image with a slightly high height (h), how can I get a low height (h1) so it will look more like what I initially request? That would be really helpful.

[Jan]

You can control the dimension of a displayed image by setting the 'Position' of the axes:

axes('Position', [0.1, 0.1, 0.8, 0.1],  ...
     'visible', 'off', 'NextPlot', 'add');
image(RGB);

Or by defining the Y-limits. The same works for the patch object.

Answer 2

[Benjamin Kraus]

You can use a patch object for this.

values = [0.5 0.3 0.2];
colors = [24 70 187; 208 118 22; 88 90 28]/255;

n = numel(values);
x = [0 cumsum(values)].*[1;1];
y = ones(2,n+1).*[1;0];

vertices = [x(:) y(:)];
faces = [1 2 4 3]+((2:2:2*n)-2)';

p = patch(...
    'Vertices',vertices,...
    'Faces',faces,...
    'FaceVertexCData', colors, ...
    'FaceColor', 'flat');

xlim([-0.1 1.1]);
ylim([-0.1 1.1]);

[Salad Box]

Thank you. That did help.