When I try to find the mean rgb for the picture below, it give the value [230,160,168]. For my information, the value for red is [255,0,0]. Is it the background influence the value?

Answer 1

[Image Analyst]

Assuming the background is all perfect white (255,255,255) you can mask it

% Extract the individual red, green, and blue color channels.
redChannel = rgbImage(:, :, 1);
greenChannel = rgbImage(:, :, 2);
blueChannel = rgbImage(:, :, 3);
% Get mask:
mask = bwareafilt(min(rgbImage, 3) == 255, 1);
% Get mean R, G, and B
meanR = mean(redChannel(mask))
meanG = mean(greenChannel(mask))
meanB = mean(blueChannel(mask))

[junnie vix]

Error using bwpropfilt Expected input number 1, BW, to be two-dimensional.

Error in bwpropfilt>parse_inputs (line 118) validateattributes(bw, {'logical'}, {'nonsparse', '2d'}, mfilename, 'BW', 1);

Error in bwpropfilt (line 57) [bw, I, attrib, p, direction, conn] = parse_inputs(varargin{:});

Error in bwareafilt (line 33) bw2 = bwpropfilt(bw, 'area', p, direction, conn);

Error in color (line 18) mask = bwareafilt(min(I, 3) == 255, 1);

There is some error occur

[Image Analyst]

Sorry - forgot that the second arg of min should be []. Try this:

rgbImage = imread('1.jpg');
% Extract the individual red, green, and blue color channels.
redChannel = rgbImage(:, :, 1);
greenChannel = rgbImage(:, :, 2);
blueChannel = rgbImage(:, :, 3);
% Get mask:
whiteMask = min(rgbImage, [], 3);
whiteMask = bwareafilt(whiteMask == 255, 1);
shirtMask = ~whiteMask;
% Get mean R, G, and B
meanR = mean(redChannel(shirtMask))
meanG = mean(greenChannel(shirtMask))
meanB = mean(blueChannel(shirtMask))

[pb]

Hi, Could you comment on the line:

whiteMask = min(rgbImage, [], 3);

I didn't quite follow what happened here. I looked up the minimum function... and it seems you're finding the minimum pixed value in all three dimensions? But then it's a white mask? Sorry, new to images, so might not that great a question

Thanks

[Walter Roberson]

If you decide that white should be only (255,255,255) exactly, and that (for example) (255,253,255) or (254,254,254) should not qualify as "white" for your particular use, then you have two ways of proceeding:

all(rgbImage==255,3)

or

min(rgbImage,[],3)==255

Because the only way that a pixel can be all 255,255,255 is if the minimum component of the pixel is 255

[Image Analyst]

min(rgbImage, [], 3); is the quirky way that min() and max() work. It's not like mean(), any(), and lots of other functions. If you want to find the min along the 3rd dimension, you need to have [] be the second argument, not 3. That's just the way it is.

[Steven Lord]

You say "quirky", I say "unambigous".

This is a bit before the start of my tenure at MathWorks, but min and max were part of MATLAB before N-dimensional arrays (N > 2) were. Both min and max can accept either one input, a matrix whose minimum or maximum values you want to compute, or two inputs, where each element of the output is the minimum of the corresponding elements of each input. To avoid an ambiguity with scalar expansion for this command:

min(A, 1)

if you want to compute the minimum of A along the first dimension you have to disambiguate by calling min with 3 inputs.

min(A, [], 1)

[Image Analyst]

Of course, they could have used the name, value pairs like many/most other functions,

minValue = mean(A, 'Dimension', 3);

but I guess they wanted to keep legacy code working and not break it.

Answer 2

[John D'Errico]

First of all, there are MANY different subtle variations of red. If every pixel in that red shirt was all the same color, you would not see any variation in the color! It would just be a block of red, in the shape of a shirt. So the color [255 0 0] is not the only red. In fact, it is true that most of the pixels in that shirt are not [255 0 0], yet they are all colors that you would describe as red.

But yes, you took the mean of the entire image. The background is white. It messes up your mean computation, if all you wanted was the mean color of the red shirt.

You will need to find only the pixels from the shirt, and then form the mean of them. Be careful, as even that tag in the collar of the shirt is not at all red. But if you take the mean of that entire area, it will subtly impact the mean, though not by a huge amount.

[junnie vix]

Thanks for your quick reply. Do you have any idea on finding the pixel of the shirt?

[Image Analyst]

Use the Color Thresholder app (on the Apps tab of the tool ribbon) and segment it in LAB color space.