Subject: Re: [geometry] rtree nearest polygon to a point
From: Adam Wulkiewicz (adam.wulkiewicz_at_[hidden])
Date: 2014-12-01 21:18:53
Mark Millman wrote:
> I know this is self evident but in none of the Delaunay triangle
> libraries I recently reviewed did they first test if the point fell in
> the previous triangle. I had to modify one so that I could pass the
> index to the previous triangle back to it. In my data set (LiDAR data
> with billions of points) this improved my searches by a hundred-fold
> or more. It's very common in spatial data for points to be relatively
> close to each other and providing an optional "hint" index can speed
> things along.
Yes, this is a valid point. However I'm guessing that this depends on
the case. The points acquired from LIDAR are often close to each other
since the environment is scanned this way that subsequent points often
hit the same or very close obstacle. I'm guessing that for some random
points the result wouldn't be the same.
Btw, is this a general advice or a suggestion how the library (e.g.
rtree) could be improved?
P.S. Top-posting is discouraged on this mailing list.
Geometry list run by mateusz at loskot.net