Masking certain RGB values for an image

Hello. I was wondering if there is a way to mask/delete certain pixels in an image if you know the range of RGB values you want to remove. Thanks for any help, Olly

Answer 1

[Jan]

Yes.

RGB  = rand(400, 200, 3);
R    = RGB(:, :, 1);
G    = RGB(:, :, 2);
B    = RGB(:, :, 3);

mask2D = (0.2 < R & R < 0.3) & ...  % [EDITED]: && -> &
         (0.1 < G & G < 0.4) & ...
         (0.6 < B & B < 0.65);
mask   = cat(3, mask2D, mask2D, mask2D);

RGB(mask) = 0;

I do not know how a pixel could be deleted without destroying the shape of the image.

[Andrei Bobrov]

Hi Jan!

Can so?

&& -> &

[Jan]

@Andrei: Thanks, fixed. I'm using a C-mex for that, which short-circuits: no test of G and B components, if R does not match, etc. I expect that such a function is included in the Image Processing Toolbox.

[Image Analyst]

An alternate way if you want to use mask2D and not create the 3D mask:

% Mask the image.

maskedRgbImage = bsxfun(@times, rgbImage, cast(mask2D, class(rgbImage)));

In the end, it doesn't make any difference though - this is just the style that Sean prefers.

[Walter Roberson]

Hmmm, which is faster, cat(3,X,X,X) or repmat(X,1,1,3) ?

Logically cat(3) should be, but I can't test at the moment.

Answer 2

[Image Analyst]

See my color segmentation demos: http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862