Deblurring using weiner filters

[cipher1729]

I want to create a sort of learning-based deblurring system using weiner filters. That is, at first i'll send an image in its true and blurred form to a system, the filter present in the system should accordingly modify its (noise) parameters according to these two images. Then, when i send a different blurred image, it should be deblurred by the filter using the earlier calculated noise parameters.
Deblurring has to be done using a weiner filter.
Does anyone have any idea how do i should go about it?

Here's my progress till now:
I first thought that i would run a loop of sorts on the first blurred image with varying gaussian and additive noise parameters on the weiner filter, and then find out which deblurred image was closest to my true image. That would set fix my filter parameters. However as it turns out , it's not very easy and there are 2 parameters which can be varied, leading to a lot of possiblities.


thanks,

[fmw42]

see my fft tutorial for IM and scripts (there is a section on FFT and cameradeblur and fftdeconvol are two different implementations of the Wiener filter.

http://www.fmwconcepts.com/imagemagick/ ... urier.html
http://www.fmwconcepts.com/imagemagick/index.php