comp.lang.idl-pvwave archive: archive

Home » Public Forums » archive » Re: Kuwahara Filter

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

Re: Kuwahara Filter [message #29412]

Fri, 15 February 2002 02:57

Jaco van Gorkom
Messages: 97
Registered: November 2000

Member

"Joshua Nipper" <nipperjc@ufl.edu> wrote in message
news:a4hcm8$piu$1@spnode25.nerdc.ufl.edu...
> Hi there, I'm trying to implement the kuwahara filter, and can't seem to
> figure out how to get around using for loops, making it very slow. Can
> anyone suggest a way to speed this up?
< code skipped >

Without being aware of the finer details of the kawahara filter, I think
that you need the variance for neighbourhoods of kernelsize around
each point in your data, and then based on the variances you find around
each input point, you choose the mean value of one of the neighbourhoods.
So how about calculating the mean values for *all* neighbourhoods first,
loopless:
means = smooth(input, kernelsize)
and then calculating the variances for all neighbourhoods, again loopless:
variances = smooth( (input - means)^2, kernelsize )
followed by some clever selection step output = means[cleverStep(variances)]
involving possibly the < operator, MIN(), SHIFT() or what may be.
If all else fails, use a loop for this last step.

Hope this helps,
Jaco

Report message to a moderator

[Message index]

		Re: Kuwahara Filter By: Jaco van Gorkom on Fri, 15 February 2002 02:57
		Re: Kuwahara Filter By: Joshua Nipper on Thu, 14 February 2002 15:15

Previous Topic:	solution of linear equations
Next Topic:	Re: 5.5 bug or a local problem?

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Sat Nov 29 08:59:50 PST 2025

Total time taken to generate the page: 1.20129 seconds