determine within cell the coordinates (rows/columns) having equal 3 numbers

How can you determine within cell 'out' the coordinates (rows/columns) having equal 3 numbers?

I would need to get a matrix that shows me the coordinates with 3 equal values.

rgb = randi(255,3,3,3,'uint8');
out = mat2cell(rgb,ones(1,size(rgb,1)),ones(1,size(rgb,2)),3);

Answer 1

[dpb]

[r,c]=find(cellfun(@(c)all(c==c(1)),out));

If there are none, the above will return [] empty results; taking off the find() will always return a logical array the size of out. What's more convenient will depend mostly on the use.

Answer 2

[Dyuman Joshi]

You don't need to convert the data to cell array to find that -

rgb = randi(4,3,3,3,'uint8')

rgb = 3×3×3 uint8 array
rgb(:,:,1) =

   4   4   1
   3   1   2
   3   3   4


rgb(:,:,2) =

   4   4   1
   3   3   3
   1   2   4


rgb(:,:,3) =

   2   1   1
   4   1   2
   1   3   2

%Method 1
[r,c]=find(rgb(:,:,2)==rgb(:,:,1) & rgb(:,:,3)==rgb(:,:,1))

r = 1

c = 3

%Method 2
[r,c]=find(all(rgb(:,:,2:3)==rgb(:,:,1),3))

r = 1

c = 3

Since the data is an unsinged integer, diff() will not be useful here, otherwise we can use this -

[r,c]=find(all(diff(rgb,1,3)==0,3));

The reason being - Subtracting a bigger number from a smaller number will result in a 0 for any unsigned integer, which will give incorrect result.

[dpb]

"You don't need to convert the data to cell array to find that - "

One presumes the cell array form is the form OP has in the application already and the above machinations are just to provide a sample dataset to illustrate...

[Dyuman Joshi]

You are certainly right in presuming that.

However, the reason I mentioned that is because, in one of their previous questions OP has asked to convert a given 3D numeric array into a cell array, which I happen to have answered and on which OP asked this same question (in continutaion) but deleted their comment afterwards to ask it as a new question.

[Alberto Acri]

Exactly. I asked a new question because I was not sure whether to ask the question in the previous discussion.

@Dyuman Joshi Could I ask, using the solution of method 1, how can I select only a range of equal values? For example in the case you showed, select only the range 1:3.

[Dyuman Joshi]

I'm sorry, I didn't understand what you said. Please elaborate on your query.

[Alberto Acri]

Suppose that

rgb = randi(4,3,3,3,'uint8')

generates:

rgb = 3×3×3 uint8 array

rgb(:,:,1) =

   4   4   1
   3   1   2
   1   3   4

rgb(:,:,2) =

   4   4   1
   3   3   2
   1   3   4

rgb(:,:,3) =

   4   1   1
   4   1   2
   1   3   2

Considering that the 'rgb' written above:

[r,c] = find(rgb(:,:,2)==rgb(:,:,1) & rgb(:,:,3)==rgb(:,:,1))

generates:

r = [1; 1; 3; 2; 3];
c = [3; 1; 1; 3; 2];
% number_repeated = [1; 4; 1; 2; 3]; % is just to show which numbers are repeated

I would like to get [r,c] only in the case of repetitions, for example, of the numbers 1 and 2.

Namely, obtain:

r = [1; 3; 3];
c = [3; 1; 2];
% number_repeated = [1; 1; 3]; % is just to show which numbers are repeated

[Dyuman Joshi]

I am a bit confused.

"% number_repeated = [1; 1; 3]; % is just to show which numbers are repeated"

But 3 is not repeated above in - "% number_repeated = [1; 4; 1; 2; 3]; % is just to show which numbers are repeated"

[Alberto Acri]

You are right, I was wrong in putting in the correct numbers.

r = [1; 3; 2];
c = [3; 1; 3];
% number_repeated = [1; 1; 2]; % is just to show which numbers are repeated

[Dyuman Joshi]

I am still not sure if I get what you are saying.

Say this is the array we are working with -

rgb(:,:,1) =[4   4   1
   3   1   2
   1   2   4];
rgb(:,:,2) =[4   4   1
   3   3   2
   1   2   4];
rgb(:,:,3) =[4   1   1
   4   1   2
   1   2   2];

rgb = uint8(rgb);

[r,c] = find(rgb(:,:,2)==rgb(:,:,1) & rgb(:,:,3)==rgb(:,:,1))

r = 5×1

1 3 3 1 2

c = 5×1

1 1 2 3 3

arr = rgb(:,:,1);

%repeated values
arr(sub2ind(size(rgb,[1 2]),r,c))

ans = 5×1

4 1 2 1 2

What should be the output here?

1 and 2 appear more than once in the above list, so do you want the [r,c] values for all the 1 and all the 2? i.e.

[3 1; 1 3] %for 1

[3 2; 2 3] %for 2

[Alberto Acri]

Exactly. However, in the following way:

r = [3; 3; 1; 2]
c = [1; 2; 3; 3]

[DGM]

I agree with @Dyuman Joshi here. This seems to all be part of a long composite thread about either selecting scattered pixels from a color image and getting a list of their color tuples, and it's morphed into a convoluted mess of seemingly unnecessary cell arrays.

Maybe it's all unrelated, and maybe it's hard to be certain from the outside perspective, but it sure smells like we're dealing with consequences of questionable decisions which need to be re-evaluated before we wind up digging a hole.

[Dyuman Joshi]

@Alberto Acri, Here's an approach -

rgb(:,:,1) =[4   4   1
   3   1   2
   1   2   4];
    rgb(:,:,2) =[4   4   1
       3   3   2
       1   2   4];
rgb(:,:,3) =[4   1   1
   4   1   2
   1   2   2];

rgb = uint8(rgb);

%find the linear indices
linx = find(rgb(:,:,2)==rgb(:,:,1) & rgb(:,:,3)==rgb(:,:,1));
arr = rgb(:,:,1);

%values
val = arr(linx)

val = 5×1

4 1 2 1 2

%Get unique values and their indices
[vec, ~, ia] = unique(val, 'stable');  % Stable keeps it in the same order

%frequency of the unique values
bin = accumarray(ia, 1);

%Get indices of non-repeating values
[~,uniqueidx] = intersect(val,vec(bin<=1));

%Get the indices of repeating values by getting the difference
repeatidx = setdiff(1:numel(val),uniqueidx);

%Convert the corresponding repeating linear indices to subscripts
[r,c] = ind2sub(size(arr), linx(repeatidx))

r = 4×1

3 3 1 2

c = 4×1

1 2 3 3

Answer 3

[C B]

I have taken dummy data instead of random inorder to get right co-ordinates, please let me know if you wanted to get same co-ordinates or different.

rgb = randi(255,3,3,3,'uint8')

rgb = 3×3×3 uint8 array
rgb(:,:,1) =

   235   202   208
   136     3    50
    20    50   245


rgb(:,:,2) =

   215    53   227
   214   246   143
   183    49   101


rgb(:,:,3) =

   216    11   108
   119   131   219
   199   236   125

out = mat2cell(rgb,ones(1,size(rgb,1)),ones(1,size(rgb,2)),3);


rgb = uint8([...
    100, 100, 100; 150, 150, 200; 255, 255, 255; 
    100, 150, 200; 200, 200, 200; 50, 50, 50;
    255, 0, 255; 0, 255, 0; 150, 150, 150]);

rgb = reshape(rgb, [3, 3, 3])

rgb = 3×3×3 uint8 array
rgb(:,:,1) =

   100   100   255
   150   200     0
   255    50   150


rgb(:,:,2) =

   100   150     0
   150   200   255
   255    50   150


rgb(:,:,3) =

   100   200   255
   200   200     0
   255    50   150

out = mat2cell(rgb, ones(1, size(rgb, 1)), ones(1, size(rgb, 2)), 3);

matForm = cellfun(@(x) numel(unique(x)), out);
[rows, cols] = find(matForm == 1);
equal_coords = [rows, cols];
disp(equal_coords)

     1     1
     3     1
     2     2
     3     2
     3     3