4 ms·
> You would be far less able to recover individual strands of hair in an image that was gaussian blurred. Frequency-domain deconvolution is frequency-domain de
by mananaysiempre 8mo ago
> You would be far less able to recover individual strands of hair in an image that was gaussian blurred.
Frequency-domain deconvolution is frequency-domain deconvolution, right? It doesn’t really matter what your kernel is.
- bmandale 8mo agoAs I explain, you can't perfectly reverse these filters because of quantizing. The more the signal is attenuated, the more information is lost when quantizing. So yes, it does matter what your kernel is.