elimination of conscutive regions (generalization: ones with zeros between)
1 visualizzazione (ultimi 30 giorni)
Mostra commenti meno recenti
I need effectively eliminate (by zeroing) the consecutive "1's" between "-1's" and start/end of column at each column of matrix A, which now can be separated by any number of zeroes. The number of consecutive "1's" between "-1's" and start/end of column is > N. This is a non-trivial generalization of my previous Question.
Again, typical size(A) = [100000,1000].
See example:
A = 1 -1 0 0 1 1 0 1 1 1 1 0 0 0 1 1 -1 0 -1 1 1 -1 0 -1 1 1 1 0 1 -1
For N = 2 the expected result is
Aclean = 0 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 0 -1 0 0 -1 0 -1 1 0 1 0 0 -1
For N = 3 the expected result is
Aclean = 1 -1 0 0 1 0 0 1 0 1 1 0 0 0 0 1 -1 0 -1 1 0 -1 0 -1 1 1 1 0 1 -1
4 Commenti
Risposta accettata
Bruno Luong
il 1 Ott 2018
A = [1 -1 0;
0 1 1;
0 1 1;
1 1 0;
0 0 1;
1 -1 0;
-1 1 1;
-1 0 -1;
1 1 1;
0 1 -1];
N = 3;
[m,n] = size(A);
Aclean = A;
for j=1:n
Aj = [-1; A(:,j); -1];
i = find(Aj == -1);
c = histc(find(Aj==1),i);
b = c <= N;
im = i(b);
ip = i([false; b(1:end-1)]);
a = accumarray(im,1,[m+2,1])-accumarray(ip,1,[m+2,1]);
mask = cumsum(a);
mask(i) = 1;
Aclean(:,j) = Aclean(:,j).*mask(2:end-1);
end
Aclean
3 Commenti
Più risposte (2)
Bruno Luong
il 2 Ott 2018
That's true, somehow it's a 1D scanning problem.
I think we are close to the limit of MATLAB can do, if faster speed is still needed, then one should go to MEX programming route instead of torturing MATLAB to squeeze out the last once of speed.
1 Commento
Vedere anche
Categorie
Scopri di più su Waveform Generation in Help Center e File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!