View individual components of NTSC (Y, I and Q) image

I'm trying to figure out how to take an image in NTSC format (Y,I,Q colorspace) and plot/view each component individually. An example of this can be found on the YIQ Wikipedia page where they show the Y,I,Q components from a picture of a barn.

If I do something like this:

img = rgb2ntsc(rgbframe);

Y = img(:,:,1); % Y channel 
I = img(:,:,2); % I channel
Q = img(:,:,3); % Q channel
a = zeros(size(img, 1), size(img, 2));
just_Y = cat(3, Y, a, a);
just_I = cat(3, a, I, a);
just_Q = cat(3, a, a, Q);
figure, imshow(just_Y), title('Y channel')
figure, imshow(just_I), title('I channel')
figure, imshow(just_Q), title('Q channel')

It doesn't work because imshow is treating the image as if it's in RGB colorspace.

I can't seem to find an 'ntsc' or 'yiq' colormap.

Answer 1

[DGM]

You need to assume some nominal Y and then convert back to RGB.

A = imread('YIQ_components.jpg'); % original barn image montage
rgbframe = A(1:size(A,1)/4,:,:); % cropped RGB portion

img = rgb2ntsc(rgbframe);
[Y I Q] = imsplit(img);

a = zeros(size(img, 1), size(img, 2));
b = 0.5*ones(size(img, 1), size(img, 2)); % assume Y = 0.5
just_Y = repmat(Y,[1 1 3]); % replicate so that it can be concatenated
just_I = ntsc2rgb(cat(3, b, I, a));
just_Q = ntsc2rgb(cat(3, b, a, Q));

% concatenate for simple viewing
imshow([im2double(rgbframe) just_Y; just_I just_Q]);

The astute observer may note that the I,Q representations here aren't quite the same as in the example. I am not sure where exactly that error came from, but considering that the colors in this example are on-axis, I'd say that the wiki example is probably the one that's wrong or using a different model. For example, the locus of colors in the wiki example of I is rotated about 13-15 degrees off-axis. The luma representations match, so the assumption of Rec 470/601 luma constants should be correct. Luma-chroma models are often incorrectly treated as synonymous. It may simply be YUV or something, which might explain the rotation direction.

In the end, it's just a visualization aid. In all likelihood, the image may have been tweaked for sake of content visibility.