dummy-link

LibSpatialIndex

A library for spatially indexing kD bounding box data (based on libspatialindex)

Readme

LibSpatialIndex

CI Coverage Status

LibSpatialIndex.jl is a julia wrapper around the C API of libspatialindex, for spatially indexing kD bounding box data. The Julia library (JLL) package is produced by the script at https://github.com/JuliaPackaging/Yggdrasil/tree/master/L/LibSpatialIndex and the resulting binary artifact is downloaded from https://github.com/JuliaBinaryWrappers/LibSpatialIndex_jll.jl.

Quick Guide

A new RTree with 2 dimensions can be created using this package as follows:

import LibSpatialIndex
rtree = LibSpatialIndex.RTree(2)

Insertion

Items can be inserted using the insert! method, where

LibSpatialIndex.insert!(rtree, 1, [0.,0.], [1.,1.])
LibSpatialIndex.insert!(rtree, 2, [0.,0.], [2.,2.])

inserts two items,

  • the first with id 1, associated with the box specified by [xmin=0.0,ymin=0.0] and [xmax=1.0,ymax=1.0].
  • the second with id 2, associated with the box specified by [xmin=0.0,ymin=0.0] and [xmax=2.0,ymax=2.0].

Queries

Thereafter, you can perform queries on the rtree using either (i) intersects(rtree, minvalues, maxvalues) for all items intersecting the box specified by minvalues and maxvalues, or (ii) knn(rtree, minvalues, maxvalues, k) for the k nearest items in rtree to the box specified by minvalues and maxvalues.

Intersection

So for instance,

LibSpatialIndex.intersects(rtree, [0.,0.],[1.,1.])

will return the vector [1,2] on the rtree constructed earlier, to indicate that items with ids 1 and 2 intersects the box specified by [xmin=0.0,ymin=0.0] and [xmax=1.0,ymax=1.0].

You can also perform queries on an individual point, so

LibSpatialIndex.intersects(rtree, [1.,1.])

will return the ids [1,2] in the rtree constructed earlier, and

LibSpatialIndex.intersects(rtree, [2.,2.])

will only return the vector [2], because item 1 does not contain the point [2,2].

k Nearest Neighbors

For knn queries,

LibSpatialIndex.knn(rtree, [2.,2.], 1)

returns the vector [2] because the item with id 2 is closest to the point [2.0, 2.0], and

sort(LibSpatialIndex.knn(rtree, [2.,2.], 2))

returns the vector [1,2]. If the value of k exceeds the number of items in the rtree, then fewer than k items will be returned, so

sort(SI.knn(rtree, [2.,2.], 3))

will return the vector [1,2].

First Commit

12/09/2017

Last Touched

3 days ago

Commits

32 commits