K-Means Color Quantization

We will learn how to use cv.kmeans for data clustering of pixels to perform color quantization.

Sources:

Contents

Theory

Color Quantization is the process of reducing number of colors in an image. One reason to do so is to reduce the memory. Sometimes, some devices may have limitation such that it can produce only limited number of colors. In those cases also, color quantization is performed. Here we use k-means clustering for color quantization.

A color image has 3 features, R,G,B. So we need to reshape the image to an array of Mx3 size (M is number of pixels in image). And after the clustering, we apply centroid values (it is also R,G,B) to all pixels, such that resulting image will have specified number of colors. And again we need to reshape it back to the shape of original image.

Code

Load color image

img = cv.imread(fullfile(mexopencv.root(),'test','test1.png'), ...
    'Color',true, 'ReduceScale',2);

reshape RGB channels into Mx3 matrix, and convert to floating-point type

Z = single(reshape(img, [], 3));
whos img Z

  Name           Size                 Bytes  Class     Attributes

  Z         106400x3                1276800  single              
  img          266x400x3             319200  uint8               

define clustering parameters

opts = {'Initialization','Random', 'Attempts',10, ...
    'Criteria',struct('type','Count+EPS', 'maxCount',10, 'epsilon',0.1)};

perform clustering with various number of clusters

K = [2 4 8];
for i=1:numel(K)
    % apply kmeans
    tic
    [labels, centers] = cv.kmeans(Z, K(i), opts{:});
    toc
    whos centers

    % Convert centroids back into 8-bit, and rebuild original image
    % (note that labels are 0-based indices)
    centers = uint8(centers);
    out = reshape(centers(labels(:)+1,:), size(img));

    % Show quantized result
    subplot(2,2,i+1), imshow(out), title(sprintf('K = %d',K(i)))
end
subplot(221), imshow(img), title('Original image')

Elapsed time is 0.222953 seconds.
  Name         Size            Bytes  Class     Attributes

  centers      2x3                24  single              

Elapsed time is 0.245025 seconds.
  Name         Size            Bytes  Class     Attributes

  centers      4x3                48  single              

Elapsed time is 0.292261 seconds.
  Name         Size            Bytes  Class     Attributes

  centers      8x3                96  single