A fascinating problem, Michael.
How's this for a idea: use a direction-constrained nearest neighbour
search. (Don't bother googling this - I just made it up!). The STRtree in
JTS provides a nearest-neighbour search made efficient by walking the
R-tree node hierarchy. It should be possible to add some further logic to
the search algorithm to reject candidates which do not lie in the compass
direction of interest. (It seems like it will be straightforward to
compute whether a candidate lies with the compass slice).
You'll have to open up the STRtree nearest neighbour code and add provision
for a custom filter to be run on each candidate as they are added to the
current search list. This should be built using a filter function
interface, to make it reusable.
On Mon, Dec 2, 2013 at 4:21 PM, Michael Bedward
<[email protected]>wrote:
> Hi Martin,
>
> The following is a general algorithm query rather than a specific JTS
> question - hope that's ok.
>
> I have a set of polygons which represent urban areas within a
> fire-prone landscape. There are about 5000 polygons varying widely in
> area, spatial extent, number of vertices and boundary complexity.
>
> I have the task of creating a regular grid of sample points within the
> surrounding landscape and, for each point, determining the distance to
> the nearest polygon within each of eight compass segments (ie. N to
> NE, NE to E...). The results only need to be expressed as categorized
> values based on a small number of distance cut-points.
>
> I've been trying to nut out something other than a brute force
> approach to this without much success so far. One of the possibly less
> brutish approaches I've thought of involves the following:
>
> 1. Split up the urban polygons by intersecting them with a regular
> lattice, create PreparedGeometry objects for the resulting parts and
> put them into an STRtree.
>
> 2. Construct a sampling template made up of polygons representing the
> compass segments, each split according to the distance cut-points
> (looking a bit like a dart board), possibly using a custom
> CoordinateSequence to have vertex coordinates calculated on the fly
> based on the centre coordinate.
>
> 3. For each sample point location:
> For each compass segment:
> For inner to outer segment part:
> Query the spatial index with the segment part envelope
> and, if any urban polygons are returned, test for
> intersection.
>
> I'm sure there must be a better way. Any suggestions or comments will
> be gratefully accepted.
>
> Michael
>
>
> ------------------------------------------------------------------------------
> Rapidly troubleshoot problems before they affect your business. Most IT
> organizations don't have a clear picture of how application performance
> affects their revenue. With AppDynamics, you get 100% visibility into your
> Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics
> Pro!
> http://pubads.g.doubleclick.net/gampad/clk?id=84349351&iu=/4140/ostg.clktrk
> _______________________________________________
> Jts-topo-suite-user mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/jts-topo-suite-user
>
------------------------------------------------------------------------------
Rapidly troubleshoot problems before they affect your business. Most IT
organizations don't have a clear picture of how application performance
affects their revenue. With AppDynamics, you get 100% visibility into your
Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics Pro!
http://pubads.g.doubleclick.net/gampad/clk?id=84349351&iu=/4140/ostg.clktrk
_______________________________________________
Jts-topo-suite-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jts-topo-suite-user
