Boyko Bantchev wrote:
>  This is much harder than finding whether a point is within a
>  polygon.  

Agreed, but these are my constraints (it's a long story).  But on the up side, 
there are very few polygons but a gazillion points.  So don't have to calculate 
bounding rectangles very often.

>  Do you assume that the polygon is non-intersecting?

Yes.

> Also, what does it mean a `maximum' rectangle?
>  With respect to its area?  In any case, the solution is not
>  unique and probably not very useful.

I was thinking maximum (shared) area, yes.  And uniqueness doesn't matter -- 
the only purpose of the rectangle is to minimize the number of times I need to 
do the full polygon containment, so even a good approximation is fine.

Basically, the approach is this:
        (1)  A new polygon is defined
        (2)  I use J to calculate the minimal circumscribed (C) and maximal 
inscribed (I) rectangles for this polygon
        (3)  I set up the following rules in the other tool:
                a.  For all points outside of C, return 0
                b.  For all points inside of I, return 1
                c.  For all points inside of C but outside of I, use J to 
determine whether the point is truly in the polygon, and return its result.

This is actually a good opportunity to show off J (the use case is pretty cool).

-Dan

-----Original Message-----
From: [email protected] 
[mailto:[email protected]] On Behalf Of Boyko Bantchev
Sent: Friday, April 16, 2010 11:18 AM
To: Programming forum
Subject: Re: [Jprogramming] Polygon containment

On 16 April 2010 18:01, Dan Bron <[email protected]> wrote:
>         (2) .............
>             the maximum inscribed rectangle in that polygon.

Dan,

This is much harder than finding whether a point is within a
polygon.  Also, what does it mean a `maximum' rectangle?
With respect to its area?  In any case, the solution is not
unique and probably not very useful.

>         (3) .................  The inputs is the
>             polygon and the points, and the output is a boolean
>             per point, which indicates whether the point is
>             "truly" in the polygon .............

Do points on the polygon's boundary count as in or out?

Do you assume that the polygon is non-intersecting?

Regards,
   Boyko
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to