The question is unclear. This seems to be what was attempted, but it doesn't make sense given the question statement.
inpict = imread('brown.jpeg');
inpict = im2double(inpict);
% you can take the ratio of RGB channels versus the sum
% but this tells us nothing about the population of brown pixels
% or the three colors themselves
channelratio = mean(inpict./sum(inpict,3),1:2);
channelratio = permute(channelratio,[1 3 2])
Instead, I think it makes more sense to find the relative area covered by each of the three brown colors.
% instead of doing this using explicit color matching
% i'm just going to use rgb2ind().
inpict = imread('brown.jpeg');
inpict = im2double(inpict);
ncolors = 4; % three browns and one white
[indpict CT] = rgb2ind(inpict,ncolors,'nodither');
% CT should represent three browns and one near-white.
% find the white tuple, so that we can ignore it.
[~,whidx] = max(sum(CT,2)); % the index of white
bridx = setdiff(1:ncolors,whidx); % the indices of browns
% find the pixels belonging to each brown color
nbrownpix = zeros(1,3);
for k = bridx
nbrownpix(k) = nnz(indpict == k-1); % sum
end
nbrownpix = nbrownpix./numel(indpict) % normalize
sum(nbrownpix) % total coverage of brown
This can always be done doing any sort of color segmentation. There are plenty of examples on the forum. I'm just going to use the same colors as before.
% we can segment the image into three masks,
% one for each shade of brown.
inpict = imread('brown.jpeg');
inpict = im2double(inpict);
CT = [0.251 0.1843 0.01569; % middle
0.8902 0.7961 0.6392; % left
0.6196 0.3255 0.01961]; % right
tol = 0.1;
ncolors = size(CT,1);
nbrownpix = zeros(1,ncolors);
for k = 1:ncolors
% reorient tuple along page axis
pickedcolor = permute(CT(k,:),[1 3 2]);
% perform box (range) selection in RGB
boxmask = all(abs(inpict-pickedcolor) <= tol,3);
% sum
nbrownpix(k) = nnz(boxmask);
end
nbrownpix = nbrownpix./numel(boxmask) % normalize
