Quote:
Originally Posted by strongmanmike
there are tell tale signs of de convolution remaining on the stars, ... I know the theory is probably sound but I always doubt the accuracy of deconvolution and how people apply it, how does the filter know to shrink a feature so that it remains the actual shape it should be rather than just get shrunk to a point with arbitrary or in fact differing shape and thus simply creating more of an illusion of higher resolution...?
|
Mike,
The theory is sound and so long as you know the point spread function it should work perfectly. In practice you have to estimate (or guess) the PSF. Also, the decon implementations in packages like PI and CCDStack can only deal with a single global PSF when in real life the PSF can vary across different parts of the image.
Still, you can get reasonable results in PI so long as you have data with good SNR and don't push it too far. It has a nice tool, DynamicPSF, which helps estimate the PSF.
Cheers,
Rick.