speedup for-loops by bsxfun?

Is any solution to speedup my following code?

Supposed I have an RGB image, for each pixel, I want to find the closest color in given palette (k colors) for quantization.

PALETTE = [c1R, c1G, c1B; ....; ckR, ckG, ckB];
for i = 1:size(img,1) 
    for j = 1:size(img,2) 
%         find the closest color from PALETTE
    end
end

Answer 1

[Walter Roberson]

rgb2ind() does that, at least for some definition of "closest". You need to be specific about how you want to compute "closest" when it comes to colour.

If the task is to quantize the colors, then

Q = reshape(PALETTE(rgb2ind(YourImage, PALETTE),:),size(YourImage));

[Stephen23]

'...at least for some definition of "closest" '

Just not a very perceptual one: https://en.wikipedia.org/wiki/Color_difference