[vtkusers] Re: [Insight-users] Generating Splats and/or Point basedmodeling

Arash Jahangir arash at vije.ca
Fri Jun 25 23:05:45 EDT 2004


Hi Luis,
I am not sure I understand you.  So let me clarify.  Yes I have a point set
and an associated image.  Think of this as a colored point cloud (xyz rgb).
My problem is that I need to convert this set to a standard point cloud
(i.e. just xyz set).  But I still want to maintain the texture values via
some form of dithering such that point density indicates the shading as well
as the shape.  A solution I have considered is to create a polygon mesh and
then apply standard dithering along the surface of every cell.   The problem
with this approach is that I will have no control over the number of points
and I'd like to be able to put a cap on the final number of points.  Equally
important, I am not convinced this is the right approach to begin with; if
the object has areas of complex shape with little texture (or light color)
how will that compare to regions that have simple geometry but rich texture?
I need a transfer function of sorts that can define point density as a
function of geometry curvature, intensity, and intensity variation.  I even
considered upsampling the point set and then modifying the error function in
quadratic decimation to take texture, geometry, and normals into account.
But if I am not mistaken, the standard algorithm would only take intensity
variation into account and not the absolute values.

any ideas?

thanks,
Arash


----- Original Message ----- 
From: "Luis Ibanez" <luis.ibanez at kitware.com>
To: "Arash Jahangir" <arash at vije.ca>
Cc: "ITK" <insight-users at itk.org>; "vtkList" <vtkusers at vtk.org>
Sent: Friday, June 25, 2004 7:42 PM
Subject: [vtkusers] Re: [Insight-users] Generating Splats and/or Point
basedmodeling


>
> Hi Arash,
>
>  From your email, it seems that you have
>
> 1) A Point Set
>
> 2) an Image (the texture)
>
> 3) a Geometrical model of an object
>
> 4) an association between (2) and (3)
>     which is what I would assume you mean
>     by "texture wrapped object"
>
> and that you want to find the geometrical
> relationships between (1) and (2+3).
>
>
> Is that a fair description ?
>
>
> Is that's the case, you could work out
> a variation of the metric used for ICP.
>
> In your variant you could take into account
> characteristics like second order derivatives
> of the intensities in the texture.
>
> You may want to look at the implementation
> of the PointSetToPointSet metric used for ICP
>
http://www.itk.org/Insight/Doxygen/html/classitk_1_1IterativeClosestPointMetric.html
>
> and to the examples in
>
>    Insight/Examples/Registration/
>              IterativeClosestPoint1.cxx
>              IterativeClosestPoint2.cxx
>              IterativeClosestPoint3.cxx
>
>
>
>    Regards,
>
>
>        Luis
>
>
>
>
> ---------------------
> Arash Jahangir wrote:
>
> > Hi all,
> > Given a set number of points and a texture wrapped object, I am looking
> > for suggestions as to how best to derive a splat or point cloud.  I
> > suspect the best solution will be an optimization between the geometry
> > and the texture.  But I can't think of how to derive the optimization
> > function between the texture and geometry.  Note that I am not really
> > concerned with the rendition (which is what most of the literature on
> > splats is about), rather I just want to have a decent point based
> > model.  Any help is much appreciated.
> >
> > thanks,
> > Arash
> >
> >
> >
> >
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > Insight-users mailing list
> > Insight-users at itk.org
> > http://www.itk.org/mailman/listinfo/insight-users
>
>
>
> _______________________________________________
> This is the private VTK discussion list.
> Please keep messages on-topic. Check the FAQ at:
<http://public.kitware.com/cgi-bin/vtkfaq>
> Follow this link to subscribe/unsubscribe:
> http://www.vtk.org/mailman/listinfo/vtkusers
>




More information about the vtkusers mailing list