Deconvolution - Lucy Richardson Algorithm

[rcolombari]

Hi,
if I am not wrong the LR algorithm, given a pixel at position (i,j) in the image, uses the surrounding pixel values (let consider, for instance a square) for calculating the probability at each iteration.

Is it possible in PI to set the maximum distance from pixel (i,j) of the surrounding pixels that will be used in the algorithm? Basically, is it possible to set the dimension of the square area of pixels around the pixel (i,j) ?

Thanks, regards.

[Carlos Milovic]

As I understand it, at each step of the algorithm the current image is operated with a low-pass version (i.e. a blured version). This filtering depends on the defined PSF. So, the expectation-maximization process depends only of the PSF and the regulatization that is performed. In PI, regularization is done with a wavelet based thresholding.
So, the only way to change the neighbourhood of pixels at each step is to change the PSF, or the regularization parameters.