Re: Fast search for a number within tolernaces
- From: Wibble <Wibble@xxxxxxxxxxxxxx>
- Date: Tue, 07 Jun 2005 07:12:16 -0400
Jonas Forssell wrote:
Gentlemen,
I have a set of three dimensional nodes - each with a position in space (x,y,z). I need to write a fast algorithm in Java to merge nodes that are close - i.e. within a specific tolerance.
Easy way: Run through the array of nodes, check each node against every other node and merge if needed. This will take an awful amount of time when the array is 500.000 nodes or larger.
Smart way:?
Thanks for your help
If all your nodes are equidistant, do you merge them into a single node?
Partition your space into a 3D matrix with each cube of a discrete size. Merge nodes that are within the same cube. Thats O(n). You can use a HashMap to model a sparse matrix. You can position the final node at an average of the component positions.
Read a book on graphics like Foley&VanDamm. Theres lots of ways to partition and filter 3D models that are more efficient and look better than node merging. .
- Follow-Ups:
- Re: Fast search for a number within tolernaces
- From: Jonas Forssell
- Re: Fast search for a number within tolernaces
- References:
- Fast search for a number within tolernaces
- From: Jonas Forssell
- Fast search for a number within tolernaces
- Prev by Date: Re: finally block does not complete normally - Eclipse
- Next by Date: Windows/Linux/Mac GUI?
- Previous by thread: Re: Fast search for a number within tolernaces
- Next by thread: Re: Fast search for a number within tolernaces
- Index(es):
Relevant Pages
|