[Insight-users] python prototyping advice

Thu Aug 18 03:41:32 EDT 2005

On Wed, 17 Aug 2005 20:46:10 +0200, Einstein, Daniel R  
<daniel.einstein at pnl.gov> wrote:

> Gaetan,
>
> I am very much interested in prototyping with python as well. I have a  
> little experience with ITK now, and have worked somewhat with the  
> InsightToolkit module (unfortunately this is something I must do in my  
> spare time). I find the wrapping uneven. I was able to do it on one of  
> my machines but the other is being a bit more difficult.
>
> Anyway, could you clarify the comment you made to Richard about ipython?  
> What exactly does the ipython patch allow you to do?

Using dir() function on an ITK object to get the list of attributes return  
a list with attributes names and a non string object. This non string  
object was breaking completion of attributes in ipython... itk was  
impossible to use completion on ITK objects.
The patch allow to use completion on ITK objects, which is very useful  
while prototyping with ITK because ITK names are very long.

Another great feature of ipython is the -qthread (or -wthread and  
-gthread) command line option. This option allows to use Qt gui without  
freezing the interperter. With that option, I have developed a small  
viewer with Qt and VTK which let me see in 3D the 3D images I have  
produced... really useful too for prototyping

>
> Also, in terms of your public directory, I see all of the classes that  
> you have wrapped. But these won't these need to be re-wrapped on our  
> local machines? Or am I missing something?

There is only sources in public repository, you have to build binaries for  
your own system, but that build is very simple: you just need to install  
cableswig and to activate ITK_CSWIG_PYTHON in cmake

All comments about wrappers are very welcome :-)

Regards,

Gaetan

>
> Thank you for your help
>
> Best Regards,
> Dan
>
>
> Daniel R Einstein, PhD
> Biological Monitoring and Modeling
> Pacific Northwest National Laboratory
> P.O. Box 999; MSIN P7-59
> Richland, WA 99352
> Tel: 509/ 376-2924
> Fax: 509/376-9064
> daniel.einstein at pnl.gov
>
>
> -----Original Message-----
> From: insight-users-bounces+daniel.einstein=pnl.gov at itk.org  
> [mailto:insight-users-bounces+daniel.einstein=pnl.gov at itk.org] On Behalf  
> Of Gaetan Lehmann
> Sent: Tuesday, August 16, 2005 11:58 PM
> To: Richard.Beare at ieee.org
> Cc: insight-users at itk.org
> Subject: Re: [Insight-users] python prototyping advice
>
>
> Hi Richard,
>
> I'm glad to see another ITK user who wants to do prototyping with python.
> I'm not the only one ! ;-) Sadely prototyping with python is not a  
> priority for ITK developers, and there is lot of things which make  
> prototyping with python difficult (the small number of wrapped class to  
> begin).
>
> You may be interested in the work I have already done on wrapper. ITK is  
> a huge library, and wrapping all classes with a large set of types would  
> produce too large executables. To increase the number of wrapped  
> classes, I decided to wrap only unsigned short and float type, in 2 and  
> 3 dimensions (US2, US3, F2, F3). This way, I have wrapped lots more  
> filters with only a small increase of executables size.
> This work is kept in sync with ITK CVS repository and is available in a  
> darcs repository at http://voxel.jouy.inra.fr/darcs/itk-mima2/
> It also include some work done on mathematical morphology filters, and a
> module for python (the module is named itk) which introduce facilities
> for prototyping, especially with ipython.
> If you are a ipython user, I advise you to checkout a fresh svn version  
> of
> ipython: it contains a patch which allow completion to work with ITK.  
> That make prototyping really easier. If you're not a ipython user, you  
> should try it quickly ;-)
>
> If you're interested in improving ITK, you should consider joining the  
> developers mailing list which is the right list to talk about that. On  
> this list, we are also talking with Benoit Regrain of what can be done  
> to make ITK more usable with python and to improve wrap process. You are  
> welcome in the discussion if you're interested in :-)
>
> Gaetan
>
> On Wed, 17 Aug 2005 08:01:07 +0200, Richard Beare  
> <richard.beare at gmail.com> wrote:
>
>> Hi,
>>
>> I'm trying to stick with prototyping using python for simpler trials.
>> At the moment this includes mimicing a few old matlab scripts. This
>> brings me to my two questions.
>>
>> Every time I've attempted to do something in python I've been slowed
>> down by missing wrappers. Some of these are easy enough to add (I hope
>> this is true - I'm waiting for the build to finish at the moment).
>> Assuming I've done this successfully, is there any way I can commit
>> these additions back to the repository? What is the protocol for this
>> type of thing?
>>
>> The other question relates to certain types of operations that are
>> often done in interactive interpreter environments like matlab or R.
>> The particular matlab script I'm porting has a step where every pixel
>> in a mask is weighted by the distance to the mask center of gravity.
>>
>> The approach in matlab is to generate coordinate images and operate on
>> them e.g
>>
>> dist = sqrt((xcoords - xcog).^2 + (ycoords - ycog).^2);
>>
>> result = mask * dist
>>
>> How do I do something equivalent in python? Is this where the numarray
>> stuff comes in?
>>
>> Thanks
>> _______________________________________________
>> Insight-users mailing list
>> Insight-users at itk.org
>> http://www.itk.org/mailman/listinfo/insight-users
>
>
>
> --
> Gaetan Lehmann <gaetan.lehmann at jouy.inra.fr>
> Tel: +33 1 34 65 29 66
> Biologie du Développement et de la Reproduction INRA de Jouy-en-Josas  
> (France)
> Web: http://voxel.jouy.inra.fr
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>

-- 
Gaetan Lehmann <gaetan.lehmann at jouy.inra.fr>
Tel: +33 1 34 65 29 66
Biologie du Développement et de la Reproduction
INRA de Jouy-en-Josas (France)
Web: http://voxel.jouy.inra.fr