How to divide a triangulation

There is a triangulation, which is two parts that are not connected, how to distinguish them?

For example, a triangulation TR, which connection matrix is TR.ConnectivityList

[3 4 5;

4 6 3;

5 3 6;

8 9 2;

9 8 2]

As we can see that 3 4 5 6 are connected and 2 8 9 are connected, how to distinguish them to two triangulations, Tr1 and TR2

I tried to use dual for loop or dual while. Although the function can be realized, but the efficiency is very low, and the actual data is thousands of points, (3D point)

Now I use undirected graph to deal with them, it is faster than for or while, but the process is very complicated,,,

Is there a simpler way? or some matlab build-in function/statement?

connection=[TR.ConnectivityList(:,[1 2]);TR.ConnectivityList(:,[2 3]);TR.ConnectivityList(:,[1 3])];
connection=unique(sort(connection,2),'rows');
G=graph(connection(:,1),connection(:,2));%node connectivity, has lot of noise points
%plot(G)
bins = conncomp(G);
for i=1:max(bins)
    p_index=find(bins==i);
    m=ismember(TR.ConnectivityList,p_index);
    t_ind=sum(m,2)>0;
    TR_sorted{i,:}=triangulation(TR.ConnectivityList(t_ind,:),TR.Points);
end

Answer 1

[Jun Matsubayashi]

I am facing the same problem as you. It may be too late to reply, but I suggest one solution. This code does not involve any for loop.

tmp=TR.ConnectivityList;
groupnumber=1;
subTR=cell(1,1);

while ~isempty(tmp)
    fprintf('Finding group #%d', groupnumber);
    poolvec=tmp(1,:);
    poolvecsize=3;
    while length(poolvecsize)==1 || poolvecsize(end)>poolvecsize(end-1)
        ind=any(ismember(tmp,poolvec),2);
        poolvec=unique([reshape(tmp(ind,:),1,[]) poolvec]);
        poolvecsize=[poolvecsize length(poolvec)];
        fprintf('.');
    end
    fprintf('\nCumulative number of faces: %d\n', sum(ind));
    subTR{groupnumber}=tmp(ind,:);
    tmp=tmp(~ind,:);
    groupnumber=groupnumber+1;
end