how does Matlab calculate the mean of whole image?

Question

I know how to take the mean of the whole image which is of size mxnx3 uint8 by using the following command

m = mean(I(:));

my understanding of this command is supposed we have a matrix

 A=[1 2 3;4 5 6; 7 8 9]; 

 mean_1=mean(A(:));

output is

 A =

 1     2     3
 4     5     6
 7     8     9


mean_1 =

       5

A color image is stored as an mxnx3 matrix where each element is the RGB value of that particular pixel (hence it’s a 3D matrix). You can consider it as three 2D matrices for red, green and blue intensities. so how mean is calculated in this case for three 2D matrices in Matlab?

When you do A(:) to any matrix of any size, it gets converted into a column vector. You know the rest. Right? — Sardar Usama
yes I know from that converted column vector we estimate mean, but my question is for a color image we have three matrices, among them which one is used or all three are used and converted into column vector or something else i don't know and that is what i want to know @SardarUsama — anil
As Sardar wrote in his comment, the colon operator (:) turns an array of any dimension into a column vector. So, the whole mxnx3 matrix is turned into a column vector. Hence, all its entries contribute to the mean. If you wanted the mean of only the red RGB data, you would need to do B=A(:, :, 1); mean_1 = mean(B(:)). — Amos Egel
@AmosEgel the third dimension (page) value is included while estimating the mean.,right? — anil
@anil Do you have any further questions or did Sardar and Amos address all of your concerns? — Dev-iL

Phil Goddard Phil Goddard · Accepted Answer · 2018-05-28T03:47:14

As has been suggested in the comments, you could create a temporary array for the R, the G and the B pages of the matrix and calculate their mean, but in the specific case of a 3D RGB matrix you're probably better of just doing,

rgb_mean = squeeze(mean(mean(A,1),2))

If you're not familiar with squeeze, it will convert the 3D 1x1x3 matrix that results from taking the means, into a 2D 1x3 vector, which is most likely what you are expecting.

how does Matlab calculate the mean of whole image?

1 Answers