moving averege of a matrix
1 vue (au cours des 30 derniers jours)
Afficher commentaires plus anciens
Hi!
a have a matrix:
%
A=[2 5 6 8 7 ;2 2 3 5 6; 1 2 3 4 2]
how could i calculate the moving averege (order 2) of the matrix horizentally, i mea the vectors a1= [2 5 6 8 7] and a2=[2 2 3 5 6] a3=[1 2 3 4 2] are indepandent .
the moving averege of each vector?
thank you
0 commentaires
Réponse acceptée
Andrei Bobrov
le 6 Mar 2013
out = conv2(A,[.5 .5],'same');
out(:,1:end-1);
2 commentaires
Daniel Shub
le 6 Mar 2013
For an A of 1e3 x 1e5, it appears that Jan's direct method is faster, but if you are not worried about the edge, then your method is faster.
Plus de réponses (2)
Jan
le 6 Mar 2013
Modifié(e) : Jan
le 7 Mar 2013
A = [2 5 6 8 7 ;2 2 3 5 6; 1 2 3 4 2];
B = (A(:, 1:4) + A(:, 2:5)) * 0.5;
This is one column shorter than the input, but how is the moving average defined at the edges?
[EDITED] A hard coded moving average for 5 neighboring elements along the rows:
[d1, d2] = size(X);
Z1 = zeros(d1, 1);
Z2 = zeros(d1, 2);
M = X + [Z1, X(:, 1:d1 - 2) + X(:, 3:d1), Z1] + ...
[Z2, X(:, 1:d1 - 4) + X(:, 5:d1), Z2];
divV(1, [1, 2, d1 - 1, d1]) = [1, 3, 3, 1];
divV(1, 3:d1 - 2) = 5;
Y = bsxfun(@rdivide, M, div);
Here the first element is not changed and the 2nd is the average of the elements 1 to 3. I do not assume that this can beat CONV.
3 commentaires
Jan
le 6 Mar 2013
Too big for what? Why is the first value unchanged and what exactly is "the first value"?
Image Analyst
le 6 Mar 2013
Moving how? In an m by n window? Or do you simply want the average of the entire row, in which case you can do this:
rowAverages = mean(A, 2);
1 commentaire
Image Analyst
le 7 Mar 2013
Rica, if you want the mean of 5 horizontal numbers in a 27 by 302,528 matrix, you can do this:
means = conv2(yourArray, [1 1 1 1 1]/5, 'valid');
Voir également
Catégories
En savoir plus sur Operators and Elementary Operations dans Help Center et File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!