[mkgmap-dev] patch for omitting nearby POIs with the same name and type

Mon Aug 24 17:42:29 BST 2015

Hi Gerd, both good points - I have fixed them. However, since posting and
further evaluating the results, I have come to the conclusion that whilst
the first attempt was an improvement, it was somewhat simplistic in its
approach. As ever, I started thinking of something simple and then realised
there was a better way to handle the problem. What I wanted to be able to do
was for some duplicate POIS just delete the duplicates, for others it would
be better to keep the POIS but delete the duplicate names (for example a set
of communications masts that have all been individually named with the same
name) and for yet others it would be better to delete the duplicate POI, but
move the remaining one to a midpoint between the two (useful for
multi-carriageway bridges to get the name between the carriageways and also
for roundabout names). Also, the distance for a POI to be considered a
duplicate needed to depend on the type of POI. A bollard 3m away from
another I would not consider to be a duplicate, but a roundabout 200m away
with the same name should be. I've therefore come up with another option
--nearby-poi-rules which specifies a list of rules about how to handle
nearby POIs.

Please find the updated patch attached.

Any suggestions for improvement gratefully received.

Regards,

Mike

From: Gerd Petermann [mailto:gpetermann_muenchen at hotmail.com] 
Sent: 20 August 2015 06:46
To: mkgmap-dev at lists.mkgmap.org.uk
Subject: Re: [mkgmap-dev] patch for omitting nearby POIs with the same name
and type

Hi Mike,

thanks for the patch.
I think the docu should mention that --max-same-poi-distance expects a 
value in metres.
Looking at the code I see one small -maybe unwanted - side effect:
If --max-same-poi-distance > 0 only duplicated named POI are removed.
When I coded the original code I thought that it is ok to remove
e.g. pillars when they appear at the same map point.

For those that like to test: I've uploaded a binary here:
 <http://files.mkgmap.org.uk/download/276/mkgmap.jar>
http://files.mkgmap.org.uk/download/276/mkgmap.jar

Gerd

  _____  

From:  <mailto:mike at tvage.co.uk> mike at tvage.co.uk
To:  <mailto:mkgmap-dev at lists.mkgmap.org.uk> mkgmap-dev at lists.mkgmap.org.uk
Date: Wed, 19 Aug 2015 23:14:53 +0100
Subject: [mkgmap-dev] patch for omitting nearby POIs with the same name and
type

Hi, please find attached a patch that adds two new options:
--max-same-poi-distance=distance and
--include-nearby-poi-types=type[,type...]. These options are used to
determine whether nearby POIS are duplicates and should be omitted from the
map, extending the existing check for duplicates at the exact same location.
They help to reduce the problems cause by OSM data frequently having an area
such as a building tagged with the same information as a point or where
multiple areas are given the same name - this is most problematic if the
--add-pois-to-areas option is used. It may be valid for some named POI types
to be found close together hence the option to omit certain types of POI
from the nearby processing. I have the distance set to 50 with bus stops and
cave entrances always included and this seems to work well for me (I don't
use the default style). If --max-same-poi-distance is omitted or set to 0,
the check works as previously. 

Please try and if OK, commit.

Thanks,

Mike

_______________________________________________ mkgmap-dev mailing list
<mailto:mkgmap-dev at lists.mkgmap.org.uk> mkgmap-dev at lists.mkgmap.org.uk
<http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev>
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20150824/04486acb/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: nearbypoi2.patch
Type: application/octet-stream
Size: 17420 bytes
Desc: not available
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20150824/04486acb/attachment-0001.obj>