Replace the indexes of duplicate values

I have a matrix B 1631x5.

Columns 2 and 3 represent X and Y coordinates respectively.

I want to identify the indexes where B(i+k,2)==B(i+j,2)&B(i+k,3)==B(i+j,3). Note that there can be more than one duplicate.

Below is the script of identifying the duplicates:

%%X coordinate
[~, indX] = unique(B(:, 2), 'rows');
% duplicate indices
duplicate_indX = setdiff(1:size(B, 1), indX);
% duplicate values
duplicate_valueX = B(duplicate_indX, 2);

%%Y coordinate
[~, indY] = unique(B(:, 3), 'rows');
% duplicate indices
duplicate_indY = setdiff(1:size(B, 1), indY);
% duplicate values
duplicate_valueY = B(duplicate_indY, 3);

%%Both coordinates
duplicate_ind=intersect(duplicate_indX,duplicate_indY);
duplicate_value = B(duplicate_ind, 2:3);

When the code is executed, we get 2 matrix: duplicate_ind(1x149) and duplicate_value(149x2).

Let's consider first 4 values of duplicate_ind as an example:

>> duplicate_ind(1:4)

ans =

      61    77   106   111

The corresponding values for these indexes are

>> duplicate_value(1:4,:)

ans =

    355.3035  176.9755
    364.7316  182.2644
    354.4987  202.1553
    350.5895  226.7602

Now I can find the original and the duplicate:

find(B(:,2)==duplicate_value(1,1))

ans =

       1
      61

>> find(B(:,2)==duplicate_value(2,1))

ans =

      57
      77

In this case, the index of the original value is 1 and the index of duplicate is 61. In other case: original: 57 and duplicate:77.

Now, I want to replace the indexes of the duplicates by the original ones. In our case 61 will be replaced by 1 (and 77 will be replaced by 57). Considering above, I want to build a matrix which has the size 1631x3 (must have the same number of rows as matrix B), where the 1st row represents the line number, 2nd and 3rd row - the coordinates of points.

The output is expected to look like:

1   1   2
2   2   3
3   3   4
...
57  57  58
...
61  1   62
...
77  57  78
78  78  79
...

Answer 1

[Yesbol]

Solved

 %%X coordinate
  [~, indX] = unique(B(:, 2), 'rows');
  % duplicate indices
  duplicate_indX = setdiff(1:size(B, 1), indX);
  % duplicate values
  duplicate_valueX = B(duplicate_indX, 2);

%%Y coordinate
[~, indY] = unique(B(:, 3), 'rows');
% duplicate indices
duplicate_indY = setdiff(1:size(B, 1), indY);
% duplicate values
duplicate_valueY = B(duplicate_indY, 3);

%%Both coordinates
duplicate_ind=intersect(duplicate_indX,duplicate_indY);
duplicate_value = B(duplicate_ind, 2:3);

indexes=zeros(3,size(duplicate_value,1));
for i=1:size(duplicate_value,1)
    if size(find(B(:,2)==duplicate_value(i,1)&B(:,3)==duplicate_value(i,2)),1)==2
        indexes(1:2,i)=find(B(:,2)==duplicate_value(i,1)&B(:,3)==duplicate_value(i,2));
    end
    if size(find(B(:,2)==duplicate_value(i,1)&B(:,3)==duplicate_value(i,2)),1)==3
        indexes(1:3,i)=find(B(:,2)==duplicate_value(i,1)&B(:,3)==duplicate_value(i,2));
    end    
end

for j=1:size(B,1)-1
    lines(j,1:2)=j;
    lines(j,3)=j+1;
end   

for j=1:size(lines,1)
    for i=1:1:size(indexes,2)
        if indexes(3,i)==0
            if lines(j,2)==indexes(2,i)
                lines(j,2)=indexes(1,i);
            end

              if lines(j,3)==indexes(2,i)
                  lines(j,3)=indexes(1,i);
              end
          end

          if indexes(3,i)~=0
              if lines(j,2)==indexes(3,i)
                  lines(j,2)=indexes(1,i);
              end

              if lines(j,3)==indexes(3,i)
                  lines(j,3)=indexes(1,i);
              end

              if lines(j,2)==indexes(2,i)
                  lines(j,2)=indexes(1,i);
              end

              if lines(j,3)==indexes(2,i)
                  lines(j,3)=indexes(1,i);
              end
          end

      end
  end

Answer 2

[RST]

Try this:

[~,ixu,ixb] = unique(B(:,2:3),'rows'); % indices of unique rows for columns 2 and 3
ixd = setdiff(1:size(B,1), ixu); % (duplicate rows) = (all rows) - (unique rows)

fprintf("There are %d duplicate points\n", numel(ixd))

% make a copy of the data
C = B;

% for each duplicate
for iixd=ixd
    % get indices of the matching rows
         % fd = find( all(B(:,2:3)==B(iixd,2:3), 2) );
    fd = find(ixb == ixb(iixd));
    % overwrite 2nd and subsequent matching rows with the first
    C(fd(2:end),:) = repmat( B(fd(1),:), ...
                             numel(fd)-1, 1);
end
clear fd, iixd, ixd, ixb, ixu