Re: best algorithm for inerpolation of the range data

From: frog (van.nevel_at_verizon.net)
Date: 03/25/05 

Date: Fri, 25 Mar 2005 06:42:32 GMT

The best method always is determined by what you expect to do with the data.

That said, I would suggest looking into implicit level set representations
(see Stan Osher, Guillermo Sapiro among others)

"Martin Leese" <please@see.Web.for.e-mail.INVALID> wrote in message
news:4227DA2C.4030606@see.Web.for.e-mail.INVALID...
> hossein wrote:
> > Hi Martin
> > In fact the reason for interpolation is to have an image for furthur
> > image processing tasks. The most important parts of the lidar (light
> > detection and ranging)data comes from the man-made objects (such as
> > buildings, cars,..) or natural ones (such as trees, bushes, ...).the
> > edges of these 3D objects are very important to be preserved for edge
> > detection processes and so on.
> > I hope this detail can help you to be a little familiar with what i am
> > interested to do.
>
> The standard algorithms for gridding irregular
> point data are (with increasing complexity):
>
> Inverse Distance Weighted (IDW)
> Triangulated Irregular Network (TIN)
> Spline
> Kriging
>
> All of these can preserve edges; many have
> parameters settings you can vary to help
> achieve this.
>
> In theory Kriging is the best, but it is not
> deterministic and requires iteration. Hence
> long execution times. People usually use IDW
> to grid LIDAR data because data volumes are
> large and IDW is fast. A really fast
> implementation of IDW is QuickGrid, visit
>
http://www.ermapper.com/downloads/download_view.aspx?PRODUCT_VERSION_ID=134
> This is a modified form of IDW.
>
> I would recommend you try QuickGrid first
> because it is (a) quick, and (b) free. If it
> doesn't preserve edges to the degree you need
> I would then suggest TIN which interpolates
> using only the nearest three points.
>
> --
> Regards,
> Martin Leese
> E-mail: please@see.Web.for.e-mail.INVALID
> Web: http://members.tripod.com/martin_leese/
>