| Re: Voxels in IDL [message #67748 is a reply to message #67666] |
Sun, 23 August 2009 08:21  |
weclifton
Messages: 12
Registered: May 2007
|
Junior Member |
|
|
On Aug 17, 4:38 am, Irim <amiri.r...@gmail.com> wrote:
> On Aug 13, 11:53 pm, Spon II <christoph.b...@gmail.com> wrote:
>
>
>
>
>
>> On Aug 12, 8:49 am, Irim <amiri.r...@gmail.com> wrote:
>
>>> Hi,
>
>>> I am new to the whole thing. Does anybody know about an implementation
>>> of voxel model in IDL?
>
>>> Cheers,
>>> Reza
>
>> Hi Reza,
>
>> I don't know what you mean by 'voxel model', but IDL is very well
>> suited to manipulating, and working with, 3- to 8-dimensional arrays
>> of various data types.
>> If you want a less general answer, you'll have to ask a more specific
>> question! :-)
>
>> Regards,
>> Chris
>
>> PShttp://www.dfanning.com/isa good place to start.
>
> Hi Chris,
>
> Thanks for the reply.
> I am evaluating IDL for the implementation of a 3D image elements of
> tree structure. I have acquired airborne Lidar data and now, I am
> thinking of converting the lidarcloudpointto some sort of 3D
> representation. IDL looks great in this sense but what about
> visualisation? What is the option to visualise the 3D arrays and how
> good is that compared to other options?
>
> Cheers,
> Reza
Reza,
You ask a good question. By the way, airborne lidar data is fun to
work with! You probably have a data set consisting of at least xyz
triplets and maybe some other attributes such as intensity, etc. ? If
this is the case and you want to display the data as voxels, which is
essentially a three dimensional histogram. Once you describe the
location of a voxel in xyz space as a bin in the 3d histogram you can
then manipulate the properties of the voxel by intensity, etc. as you
see fit.
If your issue is how to take the xyz triplets and create the 3d
histogram, I suggest you use hist_nd which is available at David
Fanning's site : http://www.dfanning.com/programs/hist_nd.pro.
But, if you have a large data set, which I suspect you do since you
are working with airborne lidar (1M+ pts?), you'll probably encounter
performance issues, depending on how sparse the data is and how you
choose to visualize the data. In that regard, there are techniques
using IDL's object graphics, specifically polylines, which can speed
this up considerably (see Rick Towler's posts on the subject in this
group).
There are also more general kd tree techniques, so a google on "point
cloud", "kd tree" may help
Bill
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
