[Numpy-discussion] Help speeding up element-wise operations for video processing

Hans Meine meine@informatik.uni-hamburg...
Fri Sep 26 04:50:56 CDT 2008


Am Mittwoch, 17. September 2008 01:43:45 schrieb Brendan Simons:
> On 16-Sep-08, at 4:50 AM, Stéfan van der Walt wrote:
> > I may be completely off base here, but you should be able to do this
> > *very* quickly using your GPU, or even just using OpenGL.  Otherwise,
> > coding it up in ctypes is easy as well (I can send you a code snippet,
> > if you need).
>
> I thought so too.  I briefly researched using the GPU, but I found
> that, surprisingly,  neither multitexturing in pyopengl, nor surface
> blitting in pygame was any faster than the numpy code I posted above.

It should be really fast and your algorithm is very suited because you need 
the input matrix(texture) multiple times.  This is very similar to the 
approach of this article:

 Random Walks for Interactive Alpha-Matting
 L. Grady, T. Schiwietz, S. Aharon (USA), and R. Westermann (Germany)
 which appeared in the Proc. VIIP 2005 (IASTED)

> That could be because I'm working on a machine with an integrated
> graphics chip.

I guess so.

-- 
Ciao, /  /
     /--/
    /  / ANS


More information about the Numpy-discussion mailing list