Jean-Marc Liotier wrote:
I have noticed that in many places, in countries in which
Google does not have significant commercial interest, even many villages
have part of their street grid mapped. But looking a little closer, this
is a partial mapping of a seemingly random subset of the grid, and none
of those streets have names. [..]
So is anyone aware of automated tracing techniques that Google might be
using ? Is automated tracing from legally available imagery something
that the Openstreetmap project should study ? Street grid detection and
automated tracing would make a nice JOSM plugin wouldn't it ?
Lakewalker does it for waterlines
(http://wiki.openstreetmap.org/wiki/JOSM/Plugins/Lakewalker) and
coastline tracing of Landsat imagery using other tools has been
experimented with
(http://www.mail-archive.com/talk@openstreetmap.org/msg12661.html).
Other people have been playing with Potrace and Autotrace
(http://wikitravel.org/en/Wikitravel_talk:How_to_draw_a_map#Automated_tracing_potrace_vs._autotrace)
But I have found no mention of street tracing automation yet.
Some other sources mentioning automatic feature extraction methods :
- Automatic extraction of road and water surface
from SPOT-5 Pan-Sharpened Image :
http://mapasia.org/2009/proceeding/ip/ma09_Panu.pdf
- Potential of manual and automatic feature extraction from high
resolution space images in mountainous urban areas :
http://www.isprs.org/proceedings/XXXVIII-1-4-7_W5/paper/TOPAN-174.pdf
- Road identification from satellite images based on a fuzzy reasoning
model :
http://www.gisdevelopment.net/proceedings/mapmiddleeast/2008/mme08_22.pdf
- Automatic road extraction from remotely sensed imagery :
http://www.icrest.missouri.edu/Projects/NASA/FeatureExtraction-Roads/index.htm
Look at the examples and you'll see patterns very similar to the
imprecise anonymous streets produced by Google. They convinced me that
Google is definitely extracting them automatically from orbital imagery.
I'm amazed that I can't find any other reference to that on the Web.
The mediocre quality of Google's results for now hint at a maturing
field, but that sort of automation is a huge leverage on available
imagery. Manual improvement of the resulting ways will still be needed,
but whoever puts those techniques to good use will gain global coverage
at very low cost while conserving precious human resources to focus them
on tasks adding even more value to the map.
___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk