3d Stereo Imaging the open source way. (OSSIM, OpenSceneGraph, GRASS)

Sampson, David dsampson at NRCan.gc.ca
Fri Oct 6 15:10:20 EDT 2006


Hey folks,

At last nights meeting interest was brought up concerning 3d stereo
imaging using open source solutions. So why not research it a bit. If
you can contribute comment please feel free.

Well I started this search at OSSIM, cause, well it made sense. For more
info on the project check out: http://www.ossim.org/OSSIM/Welcome.html

These folks are mainly interested in imagery and image maipulation
wihtin a geomatics context.

Next I searched for Open Source and various keywords with 3d imaging,
stereo imaging and the like.

Well journal articles are always exciting and boring at the same time.
But one in particular jumped out that looked like it might be a similar
application of what we discussed.

Check out  The article " Interactive Stereo Electron Microscopy Enhanced
with Virtual Reality   "  here:
http://www-vis.lbl.gov/Publications/2002/LBNL-48336-StereoEM.pdf

So if you take the principles here and replace the exlectron microscope
imagery with stereo air photos you may achieve the same benefits. Notice
how the images look like that red and blue cartoon outlines we used to
find in shredy boxes with the funky glasses. Basicaly it interlaces the
two images (form my understanding)

So what software did they use?... Well they reference a project called
OpenRM. Check out the overview of the project here:
http://www.openrm.org/overview.shtml

Now I'll have to research this feature more but they mention:
* Support for binocular analglyph and multibuffered stereo. 

Now before you download OpenRM and expect results, notice this is a
libray and API... it is NOT an application.

So what do we use to exploit the library?.... Well Check out Open Scene
Graph at: http://www.openscenegraph.org/

And is it also coincidence that OSSIM uses open scene graph? Which
brings us back to the circle.

So now you have a 3d stereo imaging library, a graphical environment and
a geomatics package.

Now how close is this to production quality.   No idea, but it would be
interesting to hear how people fare by piecing it together.

Any missing pieces of the puzzle or other alternatives would probably
help answer the original question.

Now another option is using GRASS and their stereo package. It is
unmaintained since 1998 but might be a good starting point if another
stereo project starts up. The source is still available.

Check out this small tutorial for a run through. Not sure if this is
eaxactly what we want but might be interesting.
Check it out: http://grass.itc.it/gdp/stereo-grass/index.html

Markus from GRASS used to work with stereo imagery and GRASS so he might
be a good resource:
http://grass.itc.it/pipermail/grassuser/2001-April/004577.html

There is the TINA project: http://www.tina-vision.net/


"The TINA libraries can be used in one of two ways.  They have been
designed to be 
 69 highly modular, allowing users to strip out parts of the
functionality to use in 
 70 their own code (under the terms of the licence).  However, they also
include code
 71 for interfaces to the algorithms.  This code is in the form of a set
of "tools", 
 72 each of which appears as a window on the screen, and which collect
together related
 73 functionality applicable to a certain task (loading single images,
stereo image
 74 pairs, temporal sequences or medical image volumes; stereo vision;
medical image
 75 segmentation; registration; blood flow analysis etc. etc. ).  In
order to use this
 76 interface, you must build a "toolkit", a program that includes the
tools you want 
 77 to access."


AND Finaly check out this page:
http://www.cs.cmu.edu/~cil/v-source.html

It lists a bunch of projects and software dealing with various vision
packages... a scattering of open source options.


If you have anything to add or have experience with these packages feel
free to let the group know.

Cheers and Goble Goble.
 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/ottawa_users/attachments/20061006/5566cd7f/attachment.html


More information about the Ottawa_users mailing list