[SciPy-User] Gaussian Filter

[Matthieu Brucher]

2009/9/5 Brian Thorne <hardbyte at gmail.com>:
> Thanks for the replies! I tried your code Sturla, and although it goes fast,
> it doesn't seem to be doing quite the same thing as the ndimage filter.
> If I am interpreting it right the difference (only at a few local points)
> seems to be very large.
> Here is an image showing intensity of each channel of the difference image:
> http://2.bp.blogspot.com/_lewp47C9PZI/SqIEHFH1EGI/AAAAAAAAAgU/0H50ceb8k7M/s1600-h/Screenshot2.png
> This is a plot looking at a singe row, we can see that the difference spikes
> the whole intensity range:
> http://4.bp.blogspot.com/_lewp47C9PZI/SqIEG3PcAzI/AAAAAAAAAgM/GD2z-Zx7hg8/s1600-h/Screenshot3.png
> Given that the gaussians are allowed to be slightly different, shouldn't the
> difference in the output image be very very small as well?
> Brian

One channel is usually 8bits, so 256 values. If you add a modulo, the
values will be similar.

Matthieu
-- 
Information System Engineer, Ph.D.
Website: http://matthieu-brucher.developpez.com/
Blogs: http://matt.eifelle.com and http://blog.developpez.com/?blog=92
LinkedIn: http://www.linkedin.com/in/matthieubrucher