How do I get the average of every n rows for every column in a matrix?

Hi everyone,

I have a matrix Mat1 1085x1376, which I need to find the average of every n rows for every column in the matrix. I did found a solution but for a vector, I dont't how to apply it on the matrix.

Example:

Mat1 = [1 2 3; 4 5 6; 7 8 9; 10 11 12]  % 4x3.  Required is the average of every n=2 rows for Mat1. 

Mat1 = 4×3

1 2 3 4 5 6 7 8 9 10 11 12

% which should be a  Matrix 
SOL = [2.5 3.5 4.5; 8.5 9.5 10.5] % 2x3

SOL = 2×3

2.5000 3.5000 4.5000 8.5000 9.5000 10.5000

For a vector I used to use following code

n = 335;

Disp1 = nanmean(reshape( [Vector1(:);nan(mod(-numel(Vector1),n),1)],n,[]))

I would be grateful for any help!

Answer 1

[David Hill]

a=randi(100,15,10);
b=[];
n=5;
for k=1:n:size(a,1)-n
  b=[b;mean(a(k:k+n-1,:))];
end

[amoda]

Hi David,

thanks, it works beyond that if the size of my matrix dividable by the chosen n or not.

Answer 2

[Matt J]

Mat1= [1 2 3; 4 5 6 ; 7 8 9; 10 11 12],

Mat1 = 4×3

1 2 3 4 5 6 7 8 9 10 11 12

[m,n]=size(Mat1);
rows=2;

SOL=reshape(Mat1, rows,1,[]);
SOL=mean(SOL,1,'omitnan');
SOL=reshape(SOL,[],n)

SOL = 2×3

2.5000 3.5000 4.5000 8.5000 9.5000 10.5000

[Matt J]

Or download sepblockfun

https://www.mathworks.com/matlabcentral/fileexchange/48089-separable-block-wise-operations

which does the same thing, but will automate all the reshaping and unreshaping for you:

Mat1= [1 2 3; 4 5 6 ; 7 8 9; 10 11 12];

SOL = sepblockfun(Mat1,[2,1],    @(x,d)mean(x,d,'omitnan'))

SOL = 2×3

2.5000 3.5000 4.5000 8.5000 9.5000 10.5000

[amoda]

Hello Matt,

thanks a lot your function is quite great for beginners just like me, but it has only one disadvantage, the size of my matrix should be dividable by the chosen n, which make the options a bit limited.

[Matt J]

Easy enough to pre-pad:

Mat1= [1 2 3; 4 5 6 ; 7 8 9; 10 11 12];

[m,n]=size(Mat1);
rows=3; %bin 3 rows

mc=ceil(m/rows)*rows;  %pre-padding
Mat1(m+1:mc,:)=nan

Mat1 = 6×3

1 2 3 4 5 6 7 8 9 10 11 12 NaN NaN NaN NaN NaN NaN

SOL=reshape(Mat1, rows,1,[]);
SOL=mean(SOL,1,'omitnan');
SOL=reshape(SOL,[],n)

SOL = 2×3

4 5 6 10 11 12

Answer 3

[Image Analyst]

You can do this in only 2 lines of code (not including setup and comments) if you have the Image Processing Toolbox with the blockproc function, which is specifically built for this kind of operation:

%===============================================================================================================================
% Setup: Define input matrix
Mat1 = [1 2 3; 4 5 6; 7 8 9; 10 11 12]  % 4x3.  Required is the average of every n=2 rows for Mat1. 

Mat1 = 4×3

1 2 3 4 5 6 7 8 9 10 11 12

% which should be a  Matrix 
% SOL = [2.5 3.5 4.5; 8.5 9.5 10.5] % 2x3

%===============================================================================================================================
% Define mean function
meanFilterFunction = @(theBlockStructure) mean(theBlockStructure.data(:));

%===============================================================================================================================
% Take the mean of 2 element by 1 element block.
% Output matrix is a matrix, half as tall as the input matrix, where every
% 2 by 1 block input block is a single value in the output matrix
% that is the mean of the elements in the block.
SOL = blockproc(Mat1, [2, 1], meanFilterFunction)

SOL = 2×3

2.5000 3.5000 4.5000 8.5000 9.5000 10.5000

[amoda]

Hi Image Analyst,

thank you for your respond, unfortunately I dont have it,