home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.graphics:9686 comp.graphics.explorer:207 comp.graphics.visualization:1332
- Newsgroups: comp.graphics,comp.graphics.explorer,comp.graphics.visualization
- Path: sparky!uunet!munnari.oz.au!bunyip.cc.uq.oz.au!brolga!vthrc
- From: vthrc@brolga.cc.uq.oz.au (Danny Thomas)
- Subject: (Q) calculating spatial density
- Message-ID: <vthrc.716467347@brolga>
- Keywords: spatial density algorithm statistics
- Sender: news@bunyip.cc.uq.oz.au (USENET News System)
- Organization: Prentice Centre, University of Queensland
- Date: Mon, 14 Sep 1992 10:42:27 GMT
- Lines: 49
-
-
- I've got a complex 2D shape with a set of points in it and need to
- generate a map of density of those points. I'm sure this topic must be
- well-explored but one complication is that there are holes in the shape
- too.
- The relatively small size of this data-set does however allow points
- to be snapped to a fine pixel grid and areas calculated by pixel-
- counting. A simple approach would be to:
- 1. count the number of data points in a circular radius around each grid-
- point
- 2. calculate the area of holes in that radius by counting hole-pixels
- 3. subtract that area from that of the circle
- 4. divide the number of data-points by the non-hole area to give
- a measure of the density
-
- but I would be interested to learn of "more analytical" algorithms that
- don't depend on quantizing coordinates. It would also be interesting to
- read about techniques that can handle arbitrary weighting functions. eg
- if an inverse square weighting is applied to distances from a grid point
- to each data-point, what do you do when that vector crosses over a hole?
- Perhaps more accurately, how could you estimate a contribution from the
- points that would exist if that hole wasn't there?
-
-
- The final step of converting a density grid into a contour map should be
- solvable with other tools we have.
-
- PS I've already received a pointer to the bivar module from avs.ncsc.org,
- and apart from the fact it doesn't seem to account for holes (or edges),
- we don't have AVS. We only have Macs & PCs but are planning on an SGI
- Indigo XS24 soon, which means we are particularly interested in Explorer
- (& Khoros?).
-
- because of server problems I'd appreciate an email copy of any replies
- posted to net-news. It might be worth reducing the number of groups any
- reply is sent to.
-
- cheers,
- X500: @c=AU [I'm about to update VTHRC entries, soon...]
- @o=University of Queensland
- @ou=Physiology and Pharmacology Department
- @ou=Vision Touch and Hearing
- @cn=Danny Thomas
- @rfc822Mailbox=vthrc@cc.uq.oz.au
- ^^^^^^^^^^^^^^^^^ <- return address
-
-
- PS & apologies for cross-posting, you'd think this a basic GIS function.
-
-