nearest neighbour deblurring

[Tony Collins-4]

I'm looking to implement nearest neighbour deblurring of a 3-slice stack.

The simplest form of this is fairly straight forward (average the top and bottom slices, Gaussian blur and multiply by 0.49 then subtract from mid slice) but prone to artefacts. I have found a paper by Yano and Kumazawa that describes a modified NN method which seems an improvement. (http://www.springerlink.com/content/xhg5xqb41y2jr8at)

The paper processes the images in Fourier space and at one point requires division by: 1-(OTF²) (eqn 16 in the paper) where OTF is the Fourier transform of the PSF.

Can this be done in real space rather than Fourier space? I'm using the Gaussian filter as an approximation of the PSF and this has me stumped.

Any help much appreciated.

Thanks,
Tony

Tony J. Collins, Ph.D.
McMaster Biophotonics Facility
Dept. Biochemistry and Biomedical Sciences HSC 4H21A
McMaster University, Hamilton, ON, L8N 3Z5
[phone] x28812(off.)/x26488(lab)
[hidden email]     www.macbiophotonics.ca