How to look up a smaller array in a larger array while preserving shape
13 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
Can Ozger
el 26 de Nov. de 2022
Comentada: Image Analyst
el 27 de Nov. de 2022
I have a logical array. I want to look for a 2x2 "square" of 1's in this array, and return whether this square is present in this array or not.
LargeArray= [0,0,0,0;1,0,0,0;1,0,0,0;1,1,0,0;1,1,0,0;1,1,0,0;0,0,0,0;0,0,0,0]
FindArray= [1,1;1,1]
When I use ismember, I get the Large array as the answer, but I am looking to get the answer of whether the smaller FindArray is present in the LargeArray or not in True or False so I can use it to tag my data. Thanks!
0 comentarios
Respuesta aceptada
Image Analyst
el 26 de Nov. de 2022
Editada: Image Analyst
el 26 de Nov. de 2022
A simple brute force for loop will do it:
LargeArray= [0,0,0,0;1,0,0,0;1,0,0,0;1,1,0,0;1,1,0,0;1,1,0,0;0,0,0,0;0,0,0,0]
FindArray= [1,1;1,1]
[rL, cL] = size(LargeArray);
[rt, ct] = size(FindArray);
foundIt = false;
for col = 1 : cL-ct
for row = 1 : rL-rt
subArray = LargeArray(row:row+rt-1, col:col+ct-1);
if isequal(subArray, FindArray)
foundIt = true;
fprintf('Found it at row %d, column %d.\n', row, col);
end
end
end
5 comentarios
Bruno Luong
el 27 de Nov. de 2022
Editada: Bruno Luong
el 27 de Nov. de 2022
@Image Analyst Sorry but my code is originally designed for binary data, as OP has specified. You shouldn't apply it for generic array without knowing what exactly conv2 does.
For generic array this modified code find index (but I can give also false positive, bu I'm not explain how to fix it because it's off topic
LargeArray= randi(9, 8, 4)
FindArray= LargeArray(2:3, 2:3)
c = conv2(LargeArray,rot90(FindArray,2),'valid');
% Match (upper-left) indexes in LargeArray
[row,col] = find(c==sum(FindArray.^2,'all'));
MatchIndex = table(row,col)
Image Analyst
el 27 de Nov. de 2022
OK, I just thought that when you said, in the comment to your original post, "On case the pattern array contains only 1s," and gave simplified code, that the original post would handle any numbers. But anyway, thanks for giving generalized solution that works for any numbers. 🙂
Más respuestas (2)
Bruno Luong
el 26 de Nov. de 2022
Editada: Bruno Luong
el 26 de Nov. de 2022
Use convolution to detect matching
% I modified it to make example more interesting
LargeArray= [0,0,0,0;1,0,0,0;1,0,0,0;1,1,0,0;1,1,1,0;0,1,1,0;0,0,0,0;0,0,0,0]
FindArray= [1,1;1,1]
shiftfun = @(B) 2*B-1; % transform 0/1 respectively to -1/1
c = conv2(shiftfun(LargeArray),rot90(shiftfun(FindArray),2),'valid');
% Match (upper-left) indexes in LargeArray
[row,col] = find(c==numel(FindArray));
MatchIndex = table(row,col)
1 comentario
Bruno Luong
el 26 de Nov. de 2022
On case the pattern array contains only 1s, the code can be simplified in single-line
% I modified it to make example more interesting
LargeArray= [0,0,0,0;1,0,0,0;1,0,0,0;1,1,0,0;1,1,1,0;0,1,1,0;0,0,0,0;0,0,0,0]
FindArray= ones(2,2)
% Match (upper-left) indexes in LargeArray
[row,col] = find(conv2(LargeArray,FindArray,'valid')==numel(FindArray));
MatchIndex = table(row,col)
DGM
el 26 de Nov. de 2022
Editada: DGM
el 26 de Nov. de 2022
I'm going to demonstrate a couple ways you can do this using neighborhood operations. The case of a solid 2x2 nhood is a bit of a simplified case, but these can be extended to more general binary pattern matching. For these examples, I'm going to shamelessly steal Bruno's improved test array.
If you have IPT, you can use bwlookup().
LargeArray = [0,0,0,0; 1,0,0,0; 1,0,0,0; 1,1,0,0; 1,1,1,0; 0,1,1,0; 0,0,0,0; 0,0,0,0];
nhood = [1 1; 1 1]; % any 2x2 neighborhood
f = @(x) isequal(x,nhood); % function that describes matching behavior
lut = makelut(f,2); % create LUT for a 2x2 nhood
mk = bwlookup(LargeArray,lut) % logical map of matches
[r c] = find(mk) % convert logical mask to row,col subscripts
Alternatively, you can use basic linear filters.
LargeArray = [0,0,0,0; 1,0,0,0; 1,0,0,0; 1,1,0,0; 1,1,1,0; 0,1,1,0; 0,0,0,0; 0,0,0,0];
nhood = [1 1; 1 1]; % any 2x2 neighborhood pattern
seb = 2.^([1 3; 2 4]-1); % index weighting array
mk = imfilter(double(LargeArray),seb) == sum(sum(seb.*nhood)) % logical map of matches
[r c] = find(mk) % convert logical mask to row,col subscripts
0 comentarios
Ver también
Categorías
Más información sobre Matrix Indexing en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!