Hi Michaël and Martin,
Thanks for confirming that this is a valid shape analysis technique. The
great thing about this list is getting instant access to a large depth and
breadth of algorithmic knowledge.
My current requirement is to sort a list of CAD figures into either
linestring or polygon shapefiles. (BTW, I'm not doing this in JUMP at the
present time, but in an external program, so I apologize for the off-topic
post.) The CAD figures are not composed of lines, thank goodness, so a
graph analysis is not necessary.
Without applying some kind of algorithm, the only available technique is
to check to see if the figure is closed. This caused a problem with utility
loops (i.e. water lines) showing up as polygons. Putting in an area
threshold check partially solved that problem (utility loops are large), but
caused large pavement polygons to show up as linestrings. I'm betting on
the fact that pavement polygons tend to be long and thin, and utility loops
tend to be more circular, although this has yet to be conclusively
demonstrated. However, the technique doesn't have to be fool-proof to save
time. It just has to be right about 80% of the time.
regards,
Larry
On Wed, Apr 27, 2011 at 7:31 PM, Martin Davis <mtncl...@telus.net> wrote:
> We implemented this as part of the JCS Conflation Suite (which was the
> reason JUMP was originally developed!)
>
> There's a whole body of theory around shape recognition - much of which is
> very complicated and hard to implement/use. The circularity ratio is one of
> the more usable techniques.
>
> Martin
>
>
> On 4/27/2011 2:00 PM, Larry Becker wrote:
>
> I had a need to classify polygon shapes into general categories such as
> roads and buildings, and I though it might be possible to use a "thinness"
> factor calculated using the area and perimeter. After fooling around with
> it for a while and not getting the answer, I goggled and found the following
> solution:
>
> lCircularity ratio: the ratio of the area of the shape to the area of a
> circle (the most compact shape) having the same perimeter. (area-perimeter
> ratio)
>
> M = ( 4 * Pi * Area of Polygon) / (Perimeter of Polygon)^2
>
> For a circle, the ratio is one; for a square, it is π / 4; for an
> infinitely long and narrow shape, it is zero.
>
> I found it at: http://ce.sharif.edu/~alamgir/
>
> Further searches indicate that this method of classification may be
> relatively unknown. Does anyone have knowledge in this area?
>
> regards,
>
> Larry Becker
>
>
> ------------------------------------------------------------------------------
> WhatsUp Gold - Download Free Network Management Software
> The most intuitive, comprehensive, and cost-effective network
> management toolset available today. Delivers lowest initial
> acquisition cost and overall TCO of any competing solution.
> http://p.sf.net/sfu/whatsupgold-sd
>
>
> _______________________________________________
> Jump-pilot-devel mailing
> listJump-pilot-devel@lists.sourceforge.nethttps://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
>
>
>
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 10.0.1209 / Virus Database: 1500/3600 - Release Date: 04/27/11
>
>
>
> ------------------------------------------------------------------------------
> WhatsUp Gold - Download Free Network Management Software
> The most intuitive, comprehensive, and cost-effective network
> management toolset available today. Delivers lowest initial
> acquisition cost and overall TCO of any competing solution.
> http://p.sf.net/sfu/whatsupgold-sd
> _______________________________________________
> Jump-pilot-devel mailing list
> Jump-pilot-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
>
>
------------------------------------------------------------------------------
WhatsUp Gold - Download Free Network Management Software
The most intuitive, comprehensive, and cost-effective network
management toolset available today. Delivers lowest initial
acquisition cost and overall TCO of any competing solution.
http://p.sf.net/sfu/whatsupgold-sd
_______________________________________________
Jump-pilot-devel mailing list
Jump-pilot-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
