Hi Rena J.
Victor Minor is wrong. You must iterate
recursivelly along the perimeter of the resulting convex hull looking
for maximum distance among nodes: that will be the circle
diameter.
But there is a simpler index for compactness, the known
Gravelius coeficient: G = shape's Perimeter /
This is one of the "seemingly trivial" mathematical problems and indeed much
more complex. There are several algorithms to solve this problem. In the past I
have used the algorithm of D.J. Elzinga and D.W. Hearn.
Rinus Deurloo
University of Amsterdam
The Netherlands.
__
Well, the math for the solution is way outside my tiny brain, but this
simple example will show that the suggested MBR method is incorrect.
Imagine we have a polygon which is a thin stripe 1.41 mile in length at a
diagonal orientation. The length of the sides of the MBR are then 1 mile
each. Obvio
Google for "enclosing circle". It is not as simple as Victor Minor suggests,
but involves convex hull and a few more steps.
Regards
Uffe Kousagaard
- Original Message -
From: "Rena J." <[EMAIL PROTECTED]>
To:
Sent: Wednesday, March 29, 2006 2:01 PM
Subject: [MI-L] smallest circle aro
RJ,
If you take the minimum bounding rectangle for the polygon
(maxE,maxN),(minE,minN), then calculate the larger of the two differences (maxE
- minE) , (maxW - minW), you can use that for the diameter of your circle.
That should represent the smallest circle needed to enclose the polygon.
Vi
