CGAL users discussion list

[Dov Grobgeld <>]

Here's another followup. Here is the single hole of the unbounded face of the same arrangement code from polygon_repairing:

It is also self intersecting. Again, I don't understand why I don't get two non self-intersecting holes.

On Sat, Jan 23, 2021 at 10:43 PM Dov Grobgeld <> wrote:

Sorry, I forgot to include the input dat file of the path of my previous email.

12
30 0
130 0
130 30
100 70
100 30
130 70
130 100
30 100
30 70
0 30
0 70
30 30

On Sat, Jan 23, 2021 at 10:41 PM Dov Grobgeld <> wrote:

Thanks! That helped, and indeed I'm able to use the winding number to get rid of the red loop in my previous post, which gets a negative winding number, and can easily be filtered. But I'm still stumped by another problem of arrangements.

Consider the following closed self intersecting path:

If I run the book example chapter6/polygon_repairing.png on the above path I get the following separated paths:

The green path is however still self intersecting! Why does the Arrangement return self intersecting paths in the outer_ccb? I guess I would have to access the holes of the unbounded face to get the non-intersecting outer curves. But will that work with the winding number class? I admit that I'm quite confused by this.

Thanks!

On Fri, Jan 22, 2021 at 1:59 PM Sebastien Loriot <> wrote:

I think in the arrangement book:
http://acg.cs.tau.ac.il/cgal-arrangement-book

there is some code to fix self-intersecting polygons using winding numbers.

The code of the example is also available at the URL above.

Best,

Sebastien.

On 1/21/21 6:10 PM, Dov Grobgeld ( via
cgal-discuss Mailing List) wrote:
> Hi all,
>
> In my efforts to deal with a large variety of fonts in my text to glyph
> generator, https://github.com/dov/pomelo
> <https://github.com/dov/pomelo>, have I encountered several cases of
> "bad" contours that I need to clean. One of the problems can be seen in
> the following figure:
>
> foo.png
> The input is the purple polyline on the left. My desired output are the
> blue and the green polylines on the right, while dropping the red
> polyline. My idea of how to do this was to use arrangements.After doing
> an arrangement, so I thought, the generated separated polylines, would
> keep the segment orientations of the input. Since I know the desired
> direction of the polyline, (I know whether I'm looking for an outer
> boundary or on a hole), I can drop polylines whose orientations are not
> as expected.
>
> Unfortunately this doesn't work. I found that boundaries of the faces
> that Arrangement_2 generate are all the in the same direction.
>
> Am I missing something?  How can I prune the red polyline?
>
> Here are the input coordinates of the input as a svg d path:
>
> m 457.2054,532.00482 -2.16634,-106.13373 2.48438,-79.73047
> 4.94872,-58.86624 15.90675,0.0928 -9.17579,50.94531 -7.96093,57.34766
> -3.03516,15.03515 -5.07812,10.16406 -7.16537,4.46318 -9.10807,1.23214
> 8.85783,1.10863 7.45076,4.90699 5.07032,10.11328 3.01562,14.85157
> 8.83504,74.12725 z
>
> Thanks!
>
> --
> You are currently subscribed to cgal-discuss.
> To unsubscribe or access the archives, go to
> https://sympa.inria.fr/sympa/info/cgal-discuss
>

--
You are currently subscribed to cgal-discuss.
To unsubscribe or access the archives, go to
https://sympa.inria.fr/sympa/info/cgal-discuss