How to look up a smaller array in a larger array while preserving shape
13 次查看(过去 30 天)
显示 更早的评论
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 个评论
采纳的回答
Image Analyst
2022-11-26
编辑:Image Analyst
2022-11-26
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 个评论
Bruno Luong
2022-11-27
编辑:Bruno Luong
2022-11-27
@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
2022-11-27
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. 🙂
更多回答(2 个)
Bruno Luong
2022-11-26
编辑:Bruno Luong
2022-11-26
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 个评论
Bruno Luong
2022-11-26
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
2022-11-26
编辑:DGM
2022-11-26
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 个评论
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 Matrix Indexing 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!