I have access to orthorectifed high res imagery of towns and would like to start creating 3d models of buildings - would actually prefer to automate this process as much as possible.
I do not have access to polygons of buildings with height.
Once models are created they will be viewed on GoogleEarth.
On the web I found out about a product callled SilverEye from Geotango - has since been bought by Microsoft - product info was saying SilverEye extracts 3d building models from single scene images.
Any ideas on how I can then create 3d Models from imagery using OSS tools.
I have access to orthorectifed high res imagery of towns and would
like to start creating 3d models of buildings - would actually
prefer to automate this process as much as possible. I do not have
access to polygons of buildings with height. Once models are created
they will be viewed on GoogleEarth.
On the web I found out about a product callled SilverEye from
Geotango - has since been bought by Microsoft - product info
was saying SilverEye extracts 3d building models from single scene
images.
Only from Quickbird and IKONOS, using "a rational function model (RFM)
defined with 78 rational polynomial coefficients (RPCs). The required
RPC parameters are provided with IKONOS and QuickBird imagery." [1]. As
I understand it, using these data orthorectification is performed and
DTM created. So the ortho image is an output, not an input.
I don't think it is possible to extract elevation from any
ortho-imagery. Ortho-image is as "flat" as any map.
Any ideas on how I can then create 3d Models from imagery using OSS
tools.
From stereo aerial imagery - Efoto [2]. It seems (just a brief check)
I have access to orthorectifed high res imagery of towns and would
like to start creating 3d models of buildings - would actually prefer
to automate this process as much as possible.
I do not have access to polygons of buildings with height.
you need to extract building footprints as 2D polygons then assign a
height to each.
idea:
- load imagery into GRASS.
- run a edge detection filter over imagery to automatically make polygons.
(?? I'm unsure of which module to use, other than custom r.mfilter)
- v.clean to remove small areas (false positives)
- v.digit with ortho image as backdrop for more cleaning (delete non-
buildings)
- add a "stories" column to 2D polygon vector
- give data or start as all as one column
- v.extrude as in above URL.
Once models are created they will be viewed on GoogleEarth.