[OSGeo-Discuss] Inquiry

[Pat Tressel]

(Two threads got started for this topic, so we've copying the other thread
over here.  I've also added a little reply to Sid's new post at the end.)

---------- Forwarded message ----------
From: Pat Tressel <ptressel at myuw.net>
Date: Fri, Aug 1, 2014 at 11:34 PM
Subject: Re: [OSGeo-Discuss] Inquiry: Help please!
To: "S.A. Mouti" <moutia at outlook.com>
Cc: OSGeo Discussions <discuss at lists.osgeo.org>

Hi, Sid!

I would like to develop an algorithm that uses remote geographic sensing
> data to automatically identify, highlight, and measure rooftops and
> buildings surfaces and contours using  Geospatial data. My preference is to
> overlay the results on one of the existing  map providers such as Google
> Earth/maps or Bing .
>
>
>
>  My aim is to get the following outputs from the proposed model:
>
>    - Accurately highlighted and identified rooftops on Google maps (using
>    geo sensing data, elevation? and
>    - Property Address or GPS coordinate.
>    - Surface and square footage available for solar power generation
>    including the position of the property(N-S or E-W). At the exact surface of
>    the south facing portion of the roof.
>    - Integrate sun tool in google maps to calculate shading for each
>    building.
>    - Total surface/square footage of the roof.
>
>
> I would appreciate your guidance on the following:
>
>    - Any individual developers or companies active in this area who would
>    be willing to undertake this challenge
>    - View on technical do-ability of the project…
>    - What free geospatial data is available/needed to build the model and
>    who the providers are? (I understand that  US cleared higher resolution
>    imagery for domestic )
>    - An idea about the overall cost  for such a model.
>
> Best regards,
>
> Sid
>

Just want to mention two things:

1) Building outlines are available for some locations in both commercial
maps (Google and Bing, for instance).  In OpenStreetMap, if buildings have
not been mapped for a specific area you're interested in, you might be able
to get local mappers to do it.  (Of course, the building outlines obtained
that way may not be accurate.  Many times, the building outline is
simplified from the actual building as it's only needed to indicate, "there
is / was a building here", e.g. for rescue workers looking for survivors
after a natural disaster.)

2) If you use satellite imagery (or possibly low-elevation imagery if you
have accurate info on the camera path and orientation), then the shadows
cast by buildings can be used to estimate their height.  A very brief web
search turns up a fair number of papers on this -- just one example, with
references to earlier work that may be more relevant:

http://www.asprs.org/a/publications/pers/98journal/january/1998_jan_35-44.pdf

-- Pat


On Sat, Aug 2, 2014 at 10:55 AM, S.A. Mouti <moutia at outlook.com> wrote:

> Thank you Barend and I look forward to hearing Sander’s views.
>
> I think another way I imagine this would be possible is through using high
> resolution elevation data (1 meter or better). In this case, a scanning
> algorithm to identify buildings contours and rooftop would do the trick
> quite accurately. The rest is integrating the Google sun tool and matching
> the elevation data to any map provider to get addresses… I know it’s
> doable. I just don’t have the technical capabilities to do it. Again, this
> is my non-expert opinion and there might be some technical or other
> limitations that I am not aware of. Hence my reach out.
>
> Kindest regards,
> Sid
>

Is there public / open-source elevation data with that accuracy, or is that
commercial?

Here's a nice summary of public elevation datasets:
http://vterrain.org/Elevation/global.html
Some specific public datasets and tiles:
http://asterweb.jpl.nasa.gov/gdem.asp
http://www.cgiar-csi.org/data/srtm-90m-digital-elevation-database-v4-1
http://topotools.cr.usgs.gov/gmted_viewer/
http://www.ngdc.noaa.gov/mgg/topo/gltiles.html
http://glcf.umd.edu/data/glsdem/

-- Pat
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/discuss/attachments/20140802/285fa466/attachment-0002.html>