kd-tree implementation?

I am looking for a kd-tree implementation that returns the set of k closest
neighbors from a point P and a specified region R. Something like this:


std::vector<MyNeighborType> neighbors;
tree->Search( P, R, neighbors ) ;

In the implementation that I have found its not possible to specify a
neighbor type. I need this because each neighbor has some unique information
that I must use in the following computations.

Any suggestions on where to find a kd-tree that returns user-specified
objects instead of basic vectors - location?

carl

On Nov 2, 5:15*pm, "carl" <carl@.com> wrote:
> I am looking for a kd-tree implementation that returns the set of k closest
> neighbors from a point P and a specified region R. Something like this:
>
> std::vector<MyNeighborType> neighbors;
> * tree->Search( P, R, neighbors ) ;
>
> In the implementation that I have found its not possible to specify a
> neighbor type. I need this because each neighbor has some unique information
> that I must use in the following computations.
>
> Any suggestions on where to find a kd-tree that returns *user-specified
> objects instead of basic vectors - location?

Memorize pointers to objects instead of objects, store the common base
class pointers into the vector and cast back the object with the
dynamic_cast.

This technique is often used in user interfaces where you cannot
predict which specific set of objects the user will select.

cheers
paperab

paperab