I'd like to plot color image histograms that combine color as RGB

I'd like to plot a histogram of color images that look like this

when each color band overlaps it change colors according to the RGB combination of it. Like Red-Green overlapped area produces Yellow, Green-Blue produces Cyan etc.

Thank you in advance.

Answer 1

[Adam Danz]

By setting transparency to a value less than 1, overlapping histograms will show the combined colors. However, since the histogram colors are transparent, they aren't truely red, blue, and green and therefore the color combinations won't be truely yellow and cyan.

gaus = @(x,mu,sig,amp,vo)amp*exp(-(((x-mu).^2)/(2*sig.^2)));
x = 1:100; 
g1 = gaus(x,10,5,10); 
g2 = gaus(x,28,5,12); 
g3 = gaus(x,43,11,7); 

figure()
hold on
bins = 0:100; 
h(1) = histogram('BinEdges',bins,'BinCounts',g1, 'FaceColor', 'b');
h(2) = histogram('BinEdges',bins,'BinCounts',g2, 'FaceColor', 'g');
h(3) = histogram('BinEdges',bins,'BinCounts',g3, 'FaceColor', 'r');

The transparency levels can be changed using

set(h,'FaceAlpha',.4) % 0=completely transparent, 1=opaque

Of course you could compute the individual bar positions (histcounts) and which ones overlap in order to plot the overlapping and non-overlapping regions individually. Then you could set the colors of overlapping and non-overlapping regions independently.

[Jessada Dissopa]

it's not exactly what i wanted yet. Anyway, thank you.

[Adam Danz]

Yes, my first paragraph tells you why it isn't exactly what you wanted.

My last paragraph tells you how to get exactly what you want but it will take some work.

The answer shows the best alternative.

[Tong Minh Hoa]

How to set the colors of overlapping and non-overlapping regions independently?

Can you help me?

[DGM]

Here's an example.

% a test image
inpict = imread('rgbdistRGBCMY.png');

% a parameter
nbins = 256;

% get bin counts and positions
counts = zeros(nbins,7);
[counts(:,1) centers] = imhist(inpict(:,:,1),nbins); % R
counts(:,2) = imhist(inpict(:,:,2),nbins); % G
counts(:,3) = imhist(inpict(:,:,3),nbins); % B
counts(:,4) = min(counts(:,[2 3]),[],2); % C
counts(:,5) = min(counts(:,[1 3]),[],2); % M
counts(:,6) = min(counts(:,[1 2]),[],2); % Y
counts(:,7) = min(counts(:,1:3),[],2); % W

% plot everything
hb = gobjects(7,1);
chancolors = {'r','g','b','c','m','y','w'};
hold on;
for k = 1:7
    hb(k) = bar(centers,counts(:,k),1,chancolors{k});
end

ylim([0 3000])

Personally, I think it only makes the histogram hard to read and can easily become misleading.