Re: [Jts-topo-suite-user] Newbie question about Polygonizer
Brought to you by:
dr_jts
Menu
▾
▴
Re: [Jts-topo-suite-user] Newbie question about Polygonizer
|
From: Martin D. <mb...@re...> - 2010-01-21 17:59:21
|
Well, I did say that this was NOT a general purpose technique!
Polygonizer does not use thresholds - it assumes exact noding. To do
this more generally requires a snapping/noding system which does use
tolerances. JTS currently does not have code to do this.
M
Sanjay Rana wrote:
> Okay but it looks like that I may have spoken too soon. A much simpler
> configuration failed to produce the desirable result. Here are the
> LineStrings:
>
> [0] = {LINESTRING(434980.3309478318 5724889.6955940686
> 116.323,434975.45594999986 5724891.47745 116.323)}
> [1] = {LINESTRING(434981.75349999964 5724893.5876
> 112.323,434978.90840000007 5724885.8036 112.323)}
> [2] = {LINESTRING(434978.90840000007 5724885.8036
> 112.323,434974.03340000007 5724887.5855 112.323)}
> [3] = {LINESTRING(434974.03340000007 5724887.5855
> 112.323,434976.87849999964 5724895.3694 112.323)}
> [4] = {LINESTRING(434976.87849999964 5724895.3694
> 112.323,434981.75349999964 5724893.5876 112.323)}
>
> Are there any thresholds used by the Polygonizer methods?
>
> Best,
> Sanjay.
>
>
>
>
>
> 2010/1/21 Martin Davis <mtn...@te...>:
>
>> Yes, that's exact what I did.
>>
>> You don't need to union the linestrings before collecting them into a
>> MultiLineString, since MLSes have no constraints on the geometry of the
>> lines they contain (unlike MultiPolygons).
>>
>> Sanjay Rana wrote:
>>
>>> Hello again,
>>>
>>> I just managed to replicate your results. I basically created
>>> Multilinestring using its ordinary simple constructor and then ran
>>> union on multilinestring. Polygonizer then produced the results
>>> correctly.
>>>
>>> Thanks a lot for the idea!
>>> Sanjay.
>>>
>>> 2010/1/21 Sanjay Rana <san...@gm...>:
>>>
>>>
>>>> Hi Martin,
>>>>
>>>> It does look like the correct shape.
>>>>
>>>> "One of the side effects of the union() method is that it nodes and
>>>> dissolves (merges) linework, where there is exact coincidence. So I
>>>> tried converting your lines to a MultiLineString, then unioning it
>>>> with itself. This did create a clean noded geometry, which then
>>>> polygonized correctly."
>>>>
>>>> Does the above indicate that you first union-ed the linestring to turn
>>>> them into a Multilinestring and then union-ed the multilinestring with
>>>> itself to create the full-noded geometry, which was then polygonized?
>>>>
>>>> Thanks,
>>>> Sanjay.
>>>>
>>>> 2010/1/20 Martin Davis <mtn...@te...>:
>>>>
>>>>
>>>>> I don't actually have code - I did this in the JTS TestBuilder. It's
>>>>> quite handy for running small experiments like this.
>>>>>
>>>>> The results are:
>>>>>
>>>>> GEOMETRYCOLLECTION (POLYGON ((433359.015400001 5719394.21955,
>>>>> 433361.6749999998 5719395.0363, 433365.6593000004 5719382.0622000005,
>>>>> 433362.999705575 5719381.245431844, 433362.1041884376 5719384.161509676,
>>>>> 433359.015400001 5719394.21955)),
>>>>> POLYGON ((433355.89919999987 5719385.4471, 433358.5491000004
>>>>> 5719386.2609, 433362.1041884376 5719384.161509676, 433356.7947500007
>>>>> 5719382.530999998, 433355.89919999987 5719385.4471)),
>>>>> POLYGON ((433358.5491000004 5719386.2609, 433356.3558
>>>>> 5719393.402799999, 433359.015400001 5719394.21955, 433362.1041884376
>>>>> 5719384.161509676, 433358.5491000004 5719386.2609)),
>>>>> POLYGON ((433360.3401549162 5719380.428677129, 433357.6902999999
>>>>> 5719379.6149, 433356.7947500007 5719382.530999998, 433362.1041884376
>>>>> 5719384.161509676, 433360.3401549162 5719380.428677129)),
>>>>> POLYGON ((433362.1041884376 5719384.161509676, 433362.999705575
>>>>> 5719381.245431844, 433360.3401549162 5719380.428677129,
>>>>> 433362.1041884376 5719384.161509676)))
>>>>>
>>>>>
>>>>> Sanjay Rana wrote:
>>>>>
>>>>>
>>>>>> Wonderful - thanks for the tips, so may I see the code and results please?
>>>>>>
>>>>>> 2010/1/20 Martin Davis <mtn...@te...>:
>>>>>>
>>>>>>
>>>>>>
>>>>>>> One of the side effects of the union() method is that it nodes and dissolves
>>>>>>> (merges) linework, where there is exact coincidence. So I tried converting
>>>>>>> your lines to a MultiLineString, then unioning it with itself. This did
>>>>>>> create a clean noded geometry, which then polygonized correctly.
>>>>>>>
>>>>>>> However, this is NOT a general technique for cleaning linework, since
>>>>>>> usually non-noded endpoints never lie 100% exactly on another segment. I'm
>>>>>>> a bit surprised that it worked for your case.
>>>>>>>
>>>>>>> To handle the more general case requires code to snap endpoints to nearby
>>>>>>> endpoints, and crack edges by nearby endpoints. This would require some
>>>>>>> significant development effort to provide generally usable code. There's
>>>>>>> lots of similar code around however - check out JUMP and/or JCS for
>>>>>>> examples.
>>>>>>>
>>>>>>> Sanjay Rana wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>> Martin,
>>>>>>>>
>>>>>>>> Thanks for responding. I have indeed read your FOSS4G presentation
>>>>>>>> about noding linestrings for the purpose polygonization.
>>>>>>>>
>>>>>>>> So, I did indeed union the linestring. As far I know the edges are all
>>>>>>>> snapped. It would be great if you could post a piece of example code
>>>>>>>> that takes in a List<LineString), "nodify" them and does the
>>>>>>>> polygonization.
>>>>>>>>
>>>>>>>> Many thanks!
>>>>>>>> Sanjay.
>>>>>>>>
>>>>>>>> 2010/1/20 Martin Davis <mtn...@te...>:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>> There are a couple of issues with your data which prevents it from being
>>>>>>>>> polygonized as it stands:
>>>>>>>>>
>>>>>>>>> - The lines are not noded together correctly (eg one line terminates in
>>>>>>>>> the middle of another). They need to be noded together (which will
>>>>>>>>> probably require snapping as well)
>>>>>>>>> - There is a duplicate line down the middle of the geometry
>>>>>>>>> - I haven't verified this, but there may be endpoints which are not
>>>>>>>>> exactly coincident (snapped)
>>>>>>>>>
>>>>>>>>> Polygonize requires fully noded, snapped input to work correctly. As
>>>>>>>>> you see, it does the best it can with the geometry provided, but it
>>>>>>>>> can't correct bad input.
>>>>>>>>>
>>>>>>>>> In fact, there's nothing in JTS right now which will do snapping to a
>>>>>>>>> tolerance as is required here. You might try looking at JUMP - it has
>>>>>>>>> some tools for this.
>>>>>>>>>
>>>>>>>>> Martin
>>>>>>>>>
>>>>>>>>> Sanjay Rana wrote:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>> Hi Everyone,
>>>>>>>>>>
>>>>>>>>>> I have the following 12 LineString that I would like to Polygonize:
>>>>>>>>>>
>>>>>>>>>> [0] = {LINESTRING(433356.7947500007 5719382.5309999976
>>>>>>>>>> 84.187,433362.10418843757 5719384.1615096759 84.187)}
>>>>>>>>>> [1] = {LINESTRING(433362.10418843757 5719384.1615096759
>>>>>>>>>> 84.187,433362.999705575 5719381.2454318441 84.187)}
>>>>>>>>>> [2] = {LINESTRING(433362.999705575 5719381.2454318441
>>>>>>>>>> 84.187,433359.015400001 5719394.21955 84.187)}
>>>>>>>>>> [3] = {LINESTRING(433359.015400001 5719394.21955
>>>>>>>>>> 84.187,433362.10418843757 5719384.1615096759 84.187)}
>>>>>>>>>> [4] = {LINESTRING(433357.6902999999 5719379.6149
>>>>>>>>>> 80.687,433355.89919999987 5719385.4471 80.687)}
>>>>>>>>>> [5] = {LINESTRING(433355.89919999987 5719385.4471
>>>>>>>>>> 80.687,433358.54910000041 5719386.2609 80.687)}
>>>>>>>>>> [6] = {LINESTRING(433358.54910000041 5719386.2609 80.687,433356.3558
>>>>>>>>>> 5719393.4027999993 80.687)}
>>>>>>>>>> [7] = {LINESTRING(433356.3558 5719393.4027999993
>>>>>>>>>> 80.687,433361.67499999981 5719395.0363 80.687)}
>>>>>>>>>> [8] = {LINESTRING(433361.67499999981 5719395.0363
>>>>>>>>>> 80.687,433365.65930000041 5719382.0622000005 80.687)}
>>>>>>>>>> [9] = {LINESTRING(433365.65930000041 5719382.0622000005
>>>>>>>>>> 80.687,433357.6902999999 5719379.6149 80.687)}
>>>>>>>>>> [10] = {LINESTRING(433360.34015491622
>>>>>>>>>> 5719380.4286771286,433362.10418843757 5719384.1615096759)}
>>>>>>>>>> [11] = {LINESTRING(433362.10418843757
>>>>>>>>>> 5719384.1615096759,433358.54910000041 5719386.2609)}
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> There should be five polygons in theory but I can't seem to be able to
>>>>>>>>>> achieve that.
>>>>>>>>>>
>>>>>>>>>> Any help will be much appreciated!
>>>>>>>>>>
>>>>>>>>>> Thanks,
>>>>>>>>>> Sanjay.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ------------------------------------------------------------------------
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> No virus found in this incoming message.
>>>>>>>>>> Checked by AVG - www.avg.com
>>>>>>>>>> Version: 8.5.432 / Virus Database: 271.1.1/2634 - Release Date: 01/20/10
>>>>>>>>>> 09:12:00
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>> ------------------------------------------------------------------------------
>>>>>>>>> Throughout its 18-year history, RSA Conference consistently attracts the
>>>>>>>>> world's best and brightest in the field, creating opportunities for
>>>>>>>>> Conference
>>>>>>>>> attendees to learn about information security's most important issues
>>>>>>>>> through
>>>>>>>>> interactions with peers, luminaries and emerging and established
>>>>>>>>> companies.
>>>>>>>>> http://p.sf.net/sfu/rsaconf-dev2dev
>>>>>>>>> _______________________________________________
>>>>>>>>> Jts-topo-suite-user mailing list
>>>>>>>>> Jts...@li...
>>>>>>>>> https://lists.sourceforge.net/lists/listinfo/jts-topo-suite-user
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>> ------------------------------------------------------------------------
>>>>>>>>
>>>>>>>>
>>>>>>>> No virus found in this incoming message.
>>>>>>>> Checked by AVG - www.avg.com Version: 8.5.432 / Virus Database:
>>>>>>>> 271.1.1/2634 - Release Date: 01/20/10 09:12:00
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>> ------------------------------------------------------------------------
>>>>>>
>>>>>>
>>>>>> No virus found in this incoming message.
>>>>>> Checked by AVG - www.avg.com
>>>>>> Version: 8.5.432 / Virus Database: 271.1.1/2634 - Release Date: 01/20/10 09:12:00
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> Throughout its 18-year history, RSA Conference consistently attracts the
>>>>> world's best and brightest in the field, creating opportunities for Conference
>>>>> attendees to learn about information security's most important issues through
>>>>> interactions with peers, luminaries and emerging and established companies.
>>>>> http://p.sf.net/sfu/rsaconf-dev2dev
>>>>> _______________________________________________
>>>>> Jts-topo-suite-user mailing list
>>>>> Jts...@li...
>>>>> https://lists.sourceforge.net/lists/listinfo/jts-topo-suite-user
>>>>>
>>>>>
>>>>>
>>>> --
>>>> ---News---
>>>> Journal of Location Based Services Special Issue on NeoGeography at
>>>> http://bit.ly/2f7qiu
>>>>
>>>> See the story behind the first ever demonstration of fare-based
>>>> railways travel at
>>>> http://www.steamcircus.info
>>>>
>>>> Plot Road Traffic Injury Locations within UK and USA at
>>>> http://www.road-injuries.info
>>>>
>>>> Find out about the unique Jewish Heritage in the Roman City of Bath at
>>>> http://www.devonshire-roundhouse.info
>>>>
>>>>
>>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>>
>>> No virus found in this incoming message.
>>> Checked by AVG - www.avg.com
>>> Version: 8.5.432 / Virus Database: 271.1.1/2634 - Release Date: 01/20/10 09:12:00
>>>
>>>
>>>
>> ------------------------------------------------------------------------------
>> Throughout its 18-year history, RSA Conference consistently attracts the
>> world's best and brightest in the field, creating opportunities for Conference
>> attendees to learn about information security's most important issues through
>> interactions with peers, luminaries and emerging and established companies.
>> http://p.sf.net/sfu/rsaconf-dev2dev
>> _______________________________________________
>> Jts-topo-suite-user mailing list
>> Jts...@li...
>> https://lists.sourceforge.net/lists/listinfo/jts-topo-suite-user
>>
>>
>
>
>
>
--
Martin Davis
Senior Technical Architect
Refractions Research, Inc.
(250) 383-3022
|