2D spatial index optimized for queries of multiple regions

Question

I have a search space of a large number of axis-aligned boxes. A normal spatial index, like an r-tree, will rapidly give me a list of boxes that overlap one search area.

However, I have a large number (hundreds) of potentially overlapping search areas I'd like to query all at once. In other words, I want all objects in my data structure that overlap at least one of these five hundred boxes.

Is there a data structure optimized for this kind of query?

score 0 · Answer 1 · answered May 21 '14 at 20:55

0

You can use a quadkey and a placeholder. You can create a quadkey by interleave the x-and y co-ordinate. It's use in a morton curve a.k.a z curve.

answered May 21 '14 at 20:55

Micromega

12,486
7
35
72

Can you elaborate on how this helps beyond a normal range searching data structure? – templatetypedef May 21 '14 at 20:56
You can use a placeholder in the key:Abcd*bbcc. – Micromega May 21 '14 at 20:58
I'm still not sure I see how this would work - why would a placeholder help you do range searches over multiple overlapping ranges more efficiently than in other data structures? – templatetypedef May 21 '14 at 21:00
Isn't a multiple overlapping query uses a placeholder? – Micromega May 21 '14 at 21:07
I'm not very familiar with spatial structures, but I don't see why that would be the case. For example, how would you represent the query for the rectangles [1, 1] x [2, 2] and [2, 2] x [3, 3] as a placeholder query? – templatetypedef May 21 '14 at 22:33

2D spatial index optimized for queries of multiple regions

1 Answers1