How to look up a smaller array in a larger array while preserving shape

14 ビュー (過去 30 日間)
Can Ozger 2022 年 11 月 26 日
コメント済み: Image Analyst 2022 年 11 月 27 日
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]
LargeArray = 8×4
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]
FindArray = 2×2
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!

サインインしてコメントする。

採用された回答

Image Analyst 2022 年 11 月 26 日

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]
LargeArray = 8×4
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]
FindArray = 2×2
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
Found it at row 4, column 1. Found it at row 5, column 1.
5 件のコメント3 件の古いコメントを表示3 件の古いコメントを非表示
Bruno Luong 2022 年 11 月 27 日

@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)
LargeArray = 8×4
1 3 5 5 2 6 6 5 4 3 5 6 9 7 4 8 4 2 9 3 9 1 6 1 3 2 9 2 2 5 3 4
FindArray= LargeArray(2:3, 2:3)
FindArray = 2×2
6 6 3 5
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)
MatchIndex = 1×2 table
row col ___ ___ 2 2
Image Analyst 2022 年 11 月 27 日
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. 🙂

サインインしてコメントする。

その他の回答 (2 件)

Bruno Luong 2022 年 11 月 26 日

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]
LargeArray = 8×4
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]
FindArray = 2×2
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)
MatchIndex = 2×2 table
row col ___ ___ 4 1 5 2
1 件のコメント-1 件の古いコメントを表示-1 件の古いコメントを非表示
Bruno Luong 2022 年 11 月 26 日
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]
LargeArray = 8×4
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)
FindArray = 2×2
1 1 1 1
% Match (upper-left) indexes in LargeArray
[row,col] = find(conv2(LargeArray,FindArray,'valid')==numel(FindArray));
MatchIndex = table(row,col)
MatchIndex = 2×2 table
row col ___ ___ 4 1 5 2

サインインしてコメントする。

DGM 2022 年 11 月 26 日

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
mk = 8×4
0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
[r c] = find(mk) % convert logical mask to row,col subscripts
r = 2×1
4 5
c = 2×1
1 2
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
mk = 8×4 logical array
0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
[r c] = find(mk) % convert logical mask to row,col subscripts
r = 2×1
4 5
c = 2×1
1 2

サインインしてコメントする。

カテゴリ

Help Center および File ExchangeCreating and Concatenating Matrices についてさらに検索

R2020a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by