Any 2D matrix containing any details may be passed in any channel, in any case as it is 2D array it will be considered gray scale image. Now it is up to you, if you consider the given matrix (X) as a red component and you mix it with tow black layers (components) of the same size; X+0+0 you show the details of your image as a "red scale image" . Then if you take the same matrix but you make 0+X+0 you show the same details as a "green scale image". In the same way 0+0+X you get "blue scale image".

Here you have an example..(in MATLAB)

img = imread ('peppers.png');

imshow(img)

red = img(:,:,1);

green = img(:,:,2);

blue = img(:,:,3);

z = zeros(size(img, 1), size(img, 2));

just_red = cat(3, red, z, z);

just_green = cat(3, z, green, z);

just_blue = cat(3, z, z, blue);

original_img = cat(3, red, green, blue);

imshow(just_red), title('Red channel')

figure, imshow(just_green), title('Green channel')

figure, imshow(just_blue), title('Blue channel')

figure, imshow(original_img), title('original image')

Bit of theory - the tri-stumulus (R G B) that form primary colors are encoded based on the CIE coordinates . If you look at the mapping you see uniquely identifying the primary back can be just make the value =0 , say R=0 , B=0 should numerically give you only green.

Also , the spectral responses (of the eye) is wider of green and if you look at jpeg they use chroma blue and chroma red .