[SciPy-user] Advanced Image Processing with Python

Zachary Pincus zachary.pincus@yale....
Sat Jun 21 08:53:28 CDT 2008


Hello,

> I am sure that some of you are using advanced image processing tools  
> in
> Python.
> So, according to your experience in this field, what would be the most
> complete, advanced and up-to-date Python module for this application?
>
> Apparently, VTK, OpenCV, and PIL (too basic?) are generally used in  
> this
> field.

VTK is, as the name suggests, a visualization toolkit. It has some  
image-processing features, but by-and-large, they're the kind of thing  
you'd need for 2/3D image display (isocontouring, etc.). Its sister  
project, ITK, has more of the hard-core image-processing algorithms in  
general, and in particular, lots of tools for segmentation and  
registration.

A while ago, I participated in getting some modern python wrappers for  
ITK working. I'm not sure what the state of this work is now, but  
hopefully there should be some decent wrappers. (It's a pain, because  
ITK is designed almost completely around C++ templates, which are  
instantiated at compile-time, so making a complete run-time library  
for a dynamic language is rather tricky.)

There are some recent ctypes wrappers for OpenCV, but I haven't worked  
with those. OpenCV is sort of tricky because a lot of it was designed  
to be really low-level (so you could put it on stripped down or even  
embedded systems), but it's got good stuff.

I wind up using scipy.ndimage a lot because it is a pretty convenient  
library and does a lot of the basics, and in particular, has nice (if  
slow-ish) resampling routines.

Good luck,
Zach


More information about the SciPy-user mailing list