How to compute the mean of two disjoint region ?

Suppose I have to fragment of an image J : J_out_1 et J_out_2.

J_out_1 = J(1:h,startj:i);
J_out_2 = J(1:h,k:endj);

I would like to compute the mean of the union of those two regions , is it possible ?

m_out = mean2(J_out_1 union J_out_2);

Thank you in advance

Answer 1

[Guillaume]

m_out = mean([J_out_1(:); J_out_2(:)])

would be one way to do it assuming the image has only one colour channel. If they are RGB images:

m_out = mean([reshape(J_out_1, 1, [], 3), reshape(J_out_2, 1, [], 3)])

Note that if the two regions are the same size, you could just concatenate them without any reshaping (by colon or reshape).

[Makrim]

excellent, that's what I am looking for.finally I have done it as follow :

m_out = mean2([J_out_1 , J_out_2])

Answer 2

[Image Analyst]

Why not just take the weighted mean of the two?

numerator = numel(J_out_1) * mean2(J_out_1) + numel(J_out_2) * mean2(J_out_2)
denominator = numel(J_out_1) + numel(J_out_2)
m_out = numerator / denominator

If you want, you could make a binary image and use that as a mask to extract all the pixels in just the two regions:

binaryImage = false(size(J));
binaryImage(1:h,startj:i) = true;
binaryImage(1:h,k:endj) = true;
m_out = mean(J(binaryImage))

[Makrim]

that's another way, Yes , thank you. by the way the idea of using Logical is also brilliant.

[Image Analyst]

You're welcome. Those ways will also work even if the two subimages don't have the same number of rows. So "h" could be different for each image and they would still work.