In recent years, more and more retouchers have been entering the field without a strong background in traditional 'image processing'. There's nothing wrong with this, as much of what is involved with that isn't applicable, but I'm curious as to what areas of traditional image processing and signal analysis can be applied effectively to retouching, yet aren't in widespread use today.

One inobvious one is frequency splitting. Sean Baker started a (now quite long) thread on this (HighPass Sucks (+ solution)), and it's almost certainly not the only applicable technique.

Deconvolution can be used to sharpen images without the artifacts inherent to USM and similar. Various edge detection techniques can be used to create edge masks. FFT can be used to reduce/eliminate moire patterns.

What else?

Do you have an example of a good edge detection technique? I've been trying to improve my extractions.