summing along 3rd dimension every 10 frames

4 views (last 30 days)
Lucrezia Cester on 14 Nov 2019
Commented: the cyclist on 14 Nov 2019
Hello,
I have a matrix of 32x32x2500.
I would like to sum it along the third dimension so to have a matrix of 32x32x250 so that along the third dimesion i have summed 10 paged together thus increasing my SNR.
I have tried with something like
sumby10 = zeros(32,32,2500);
for a = 1:11:2500
sumby10(:,:,a) = sum(micron_59frames(:,:,a:a+9),3);
end
but this does not work as it return a matrix of 32x32x2500.

Philippe Lebel on 14 Nov 2019
I would do:
sumby10 = zeros(32,32,250);
for a = 1:250
sumby10(:,:,a) = sum(micron_59frames(:,:,(a-1)*10+1:(a)*10),3);
end

Stephan on 14 Nov 2019
Edited: Stephan on 14 Nov 2019
B = sum(reshape(A,32,32,10,[]),3);

the cyclist on 14 Nov 2019
Edited: the cyclist on 14 Nov 2019
Stephan provided a slick one-liner, but I think it might also be useful for you to see what mistakes you made. Here is a corrected version of your original code:
sumby10 = zeros(32,32,250);
for a = 1:250
idx = (a-1)*10+1:a*10;
sumby10(:,:,a) = sum(micron_59frames(:,:,idx),3);
end
Note a couple things that are different:
• I preallocated sumby10 to be only length 250 in dimension 3. Your preallocation guaranteed that your result would be the full 2500 in length
• I re-did the loop index. You were using the loop index "a" to write into "slices" 1,11,22,33 etc of the sumby10 matrix. You were leaving the in-between rows equal to zero -- but they were there.
• You were confusing the indexing of the old matrix (needed to span all 2500 rows) with indexing of the new matrix (only needs to span 250 rows)
the cyclist on 14 Nov 2019
¯\_(ツ)_/¯