CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

On 08/29/2014 01:14 AM, David Scriven wrote:
> Given a set, D, of (x, y, z) triplets in Euclidean 3 space, I want
> identify groups K1..n that have the following simple characteristic:
>
> For every tj that is a member of Ki  there exists a tm (m != j) s.t.
> Euclidean dist(tj, tm)  < value.
>
> Looking through CGAL it seemed that the NN incremental search would help
> me identify these groups and I modified the example given:
> ........
> int valsq = value*value;
> // A functor that returns true, iff the distance of a 3D point is
> greater than value^2
> struct X_gt_valsq {
> bool operator()(const NN_iterator& it) { return ((*it).second > valsq); }
> };
>
> typedef CGAL::Filter_iterator<NN_iterator, X_gt_valsq> NN_gtvalsq_iterator;
>
> std::list<Point_3> coords;
> ......// add points.....
> Tree tree(coords.begin(), coords.end());
> Point_3 query(x,y,z); // some member of coords
> NN_incremental_search NN(tree, query);
> NN_gtvalsq_iterator it(NN.end(), X_gt_valsq(), NN.begin()),
> end(NN.end(), X_gt_valsq());
> ......
> while(it!=end)
> {
>     std::cout << (*it).first << " at squared distance = " <<
> (*it).second << std::endl;
>     it++;
> }
>
> which works, but I can't see how to use the information from the
> iterator it (which is an input_iterator) to identify which members of
> coords are with a NN distance of < valsq so that I can delete them from
> the coords dataset to start the next search.  Any advice?
First, points are duplicated inside the search structure so you'll need 
to create a new search structure.

If you want to relate to the original points, you can use for example an 
integer as point type in the search structure. like in this example:

http://doc.cgal.org/latest/Spatial_searching/index.html#title11

Sebastien.