Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!decwrl!labrea!aurora!eos!ames!hao!gatech!ncsuvx!mcnc!unccvax!jow
From: jow@unccvax.UUCP (Jim Wiley)
Newsgroups: comp.graphics
Subject: Re: COMPLICATED PROBLEM; ONLY INTELLIGENT PEOPLE SHOULD READ
Message-ID: <917@unccvax.UUCP>
Date: 26 Feb 88 07:38:35 GMT
References: <971@ut-emx.UUCP> <210@geza.SW.MCC.COM> <21223@bbn.COM>
Organization: Univ. of NC at Charlotte, Charlotte, NC
Lines: 15

In article <21223@bbn.COM>, cosell@bbn.com (Bernie Cosell) writes:
>   that comes to mind quickly for arbitrary (!!) regions is to find a point
>   *known* to be outside the region and then take the line segment from the
>   point-being-tested to the known-outside-point.  Now check how many
>   region-boundary segments this line segment intersects.  If the number is
>   odd, then the point is inside the region, if even it is outside the region.
>   (if you handle "intesection" properly, you can just pick an endpoint of one
>   of the region-boundary lines as the known-outside point).  also note that

Intersection is not good enough.  A line drawn between to outside points
can be tangent to the region and "intersect" at only one point thus
producing a false inside indication.  The number of times the line
segment crosses the region boundry must be counted.

Jim Wiley