MGH_POLY_INSIDE
Name
MGH_POLY_INSIDE
Purpose
Determine whether a point or set of points is inside a polygon.
Calling Sequence
result = mgh_poly_inside(x, y, xp, yp)
result = mgh_poly_inside(x, y, xyp)
Positional Parameters
x, y (input, numeric scalar or array
X, Y position(s) defining the point(s) to be tested.
xp, yp (input, numeric vector)
xyp (input, numeric array)
The polygons vertices as 2 separate vectors (xp, yp) OR a single
[2,n] array.
Keyword Parameters
EDGE (input, switch)
Set this keyword to accept edge (& vertex) points as
"inside". Default is to reject them.
NAN (input, switch)
Set this keyword to specify that all points for which X or Y is
not finite (eg Nan, Inf) are to return 0. Default is to process
non-finite points, which leads to floating point errors and an
undefined result for that point.
Return Value
The function returns an array of the same shape as X. Each element
is 0 if the point is outside the polygon, 1 if it is inside the polygon.
Procedure
This routine calculates the displacement vectors from each point
to all the vertices of the polygon and then takes angles between
each pair of successive vectors. The sum of the angles is zero for
a point outside the polygon, and +/- 2*pi for a point inside. A
point on an edge will have one such angle equal to +/- pi. Points
on a vertex have a zero displacement vector.
References
Note that the question of how to determine whether a point is
inside or outside a polygon was discussed on comp.lang.idl-pvwave
in October 1999. The following is quoted from a post by Randall
Frank <[email protected]>:
I would suggest you read the Graphics FAQ on this issue and
also check Graphics Gem (I think volume 1) for a more detailed
explanation of this problem. The upshot is that there really
are three core methods and many variants. In general, you can
sum angles, sum signed areas or clip a line. There are good
code examples of all these approaches on the net which can be
coded into IDL very quickly. It also depends on how you
intend to use the function. If, you are going to repeatedly
test many points, you are better off using one of the sorted
variants of the line clipping techniques. In general, the
line clipping techniques are the fastest on the average, but
have poor worst case performance without the sorting overhead.
The angle sum is one of the slowest methods unless you can get
creative and avoid the transcendentals (and you can). The
area sum approach generally falls in between. In IDL code, I
believe you can vectorize the latter with some setup overhead,
making it the fastest for .pro code when testing multiple
points with one point per call.
Further Resources
- "Misc Notes - WR Franklin",
http://www.ecse.rpi.edu/Homepages/wrf/misc.html: includes a
reference (broken @ Jul 2001) to his point-in-polygon code.
- Comp.graphics.algorithms FAQ,
http://www.faqs.org/faqs/graphics/algorithms-faq/: See subject
2.03
See Also
MGH_PNPOLY, which implements a line-clipping technique and is much
faster.
To Do
Reduce copying of input data by segregating the calculations into a
separate function.
###########################################################################
Copyright (c) 1995-2015 NIWA:
http://www.niwa.co.nz/
Licensed under the MIT open source license:
http://www.opensource.org/licenses/mit-license.php
###########################################################################
Modification History
Mark Hadfield, 1995-06:
Written based on ideas in MATLAB routine INSIDE.M in the WHOI
Oceanography Toolbox v1.4 (R. Pawlowicz, 14 Mar 94,
[email protected]).
Mark Hadfield, 2000-12:
Updated.
Mark Hadfield, 2001-07:
Changed argument order: polygon vertices are now before test
position(s).
Mark Hadfield, 2009-09:
Changed argument order back. Polygon can now be entered as a
[2,n] array.
Mark Hadfield, 2015-02:
Updated source code format.