Re: Question on ITK per pixel operation
Charl Botha
Let's do devide questions also on the DeVIDE mailing list, that's good
for everyone!
To convert from itk image to numpy array, try the following in an
introspection window:
import itk
my_numpy_array = itk.PyBuffer[itk.Image.F3].GetArrayFromImage(your_itk_img)
(this should work on Linux 64bit DeVIDE at the least and probably
Linux 32bit. I just noticed that the Windows version for some or other
reason doesn't ship with PyBuffer :(
You *can* access voxels directly:
if3.SetPixel(itk.Index[3]((0,0,0)), 20)
(my memory is rusty, I needed a few minutes of Python introspection to
discover this)
On Sun, Apr 22, 2012 at 19:00, Christian Kehl <dj.k...@googlemail.com> wrote:
> In classical C++ style ITK, I got that it can be done by direct access - but
> I haven't been able to map that to python. Another way there is iterators,
> but we don't have such things in python ...
>
> A way would be - apart from installing SimpleITK or even bigger stuff - to
> get the ITK image into a python array (numpy/scipy - great idea Francois).
> Now, for that idea google puts out answers to itk.pybuffer - like these page
> here:
>
> http://paulnovo.us/wrapitktutorial
> https://medipy.googlecode.com/hg-history/5b17b5c542ddc73e57cb111fcd390b934aa936cb/lib/itk/numpy_bridge.py
>
> I had a look into DeVIDE by opening a python console there - pybuffer is
> missing, no valid object there.
>
> Does anybody have an idea how to solve that ? Anybody else of you already
> stoof in front of the problem ?