Re: [Boost-users] Interprenting nearest's neighbor's output

Hi Georgios, Georgios Samaras wrote:

I am running the code found here http://www.boost.org/doc/libs/1_54_0_beta1/libs/geometry/doc/html/geometry/s...

(not the intersect query, only the nn).

The output is this: " knn query point: POINT(0 0) knn query result: POLYGON((4 4,4 4.5,4.5 4.5,4.5 4,4 4)) - 4 POLYGON((3 3,3 3.5,3.5 3.5,3.5 3,3 3)) - 3 POLYGON((2 2,2 2.5,2.5 2.5,2.5 2,2 2)) - 2 POLYGON((0 0,0 0.5,0.5 0.5,0.5 0,0 0)) - 0 POLYGON((1 1,1 1.5,1.5 1.5,1.5 1,1 1)) - 1 "

What are the integers on the right? Indices of the polygons? Actually is not really clear to me what are we inserting in the tree..I tried finding some documentation on these methods, but I only found info on how to call them...

In the rtree we're storing values defined in the line: typedef std::pair value; so it's a pair of a Box and some number. This is the classic approach where in the spatial index you store AABB (Axis-Aligned Bounding Box) of some geometry and an ID of this geometry. So why Polygons are printed when you're storing Boxes? Those lines are generated by this code: std::cout << bg::wkt<box>(v.first) << " -" << v.second << std::endl; First the Box is printed in WKT format, then " - " and the ID. But WKT standard (http://en.wikipedia.org/wiki/Well-known_text) doesn't define how Boxes should be described. Boost.Geometry uses WKT Polygons for this. That's just a text output.

What I want, is simply inserting some points in the tree and then search for nn of a query point.

If you don't need any additional data, instead of std::pair just use point type from the same example. Regards, Adam