slicing an M-by-N-by-3 matrix using an M-by-N matrix of logical values

Hi everybody

I have an RGB image as a MxNx3 matrix.

I would like to slice it using an MxN matrix of logic values (one for each pixel) and add a 1x1x3 value to those pixels. I'm trying to a clean way to do it.

In other words, I'm trying to do something like this

A = rand(200,200,3);
i = rand(200,200) > 0.5;
C = cat(3,0.3,0.4,0.2);

A(i) = 0.5*A(i,:) + 0.5*C; %doesn't work!

The only ways I found is either "unpacking" each MxN submatrix to a new variable like this

R = A(:,:,1);
G = A(:,:,2);
B = A(:,:,3);
R(i) = 0.5*R(i) + 0.5*C(1);
G(i) = 0.5*G(i) + 0.5*C(2);
B(i) = 0.5*B(i) + 0.5*C(3);

Anyone got a better idea?

Answer 1

[Walter Roberson]

A = A + repmat(i, [1, 1, 3]) .* repmat(C, [size(A,1), size(A,2), 1]);

Since R2016b this can be expressed as

 A = A + i .* C;

[Luca Amerio]

The first one is nice, but the second one is great!

Nice one, thank you!

[Walter Roberson]

dpb's Answer takes into account the selective 1/2 where mine does not. The fix would be

A = A + (C - A)/2 *.i;

Answer 2

[dpb]

A=0.5*(A.*i+C.*i);

[Walter Roberson]

Note that this requires R2016b or later due to differences in the sizes of the objects.