Re: Accurate edge detection?

> I haven't lost interest.  I don't understand illywacker's sarcasm of
> "true researcher."  It would make more sense if he said "pure
> researcher."  Commercial applied research that cannot be published
> because of its proprietary nature is still true research but not pure
> research.  On the other hand, most papers published in electrical
> engineering journals represent pure research in the sense that their
> findings have no foreseeable practical use.  For engineers, that is
> immoral hypocrisy that deserves sarcasm.  And, IEEE typically charges
> $35 for you to buy a paper and then discover it is useless.  You can't
> judge a paper by its abstract anymore than judge a book by its cover.

Well, if so, what d'you think about my attempt to revert
"irrevertible" transform? Probably, there was my sarcasm, not
illywacker's one. It was my own experience of working with true-pure
researchers. I do think that isn't a proper place to tell stories
concerning my own live, but if t has some interest for you, I can send
some words what I've just sent to  illywacker. But, let's return to
our images. I've tryed to revert simple rectangular filter and was
really surprized, that this one can be reverted too with low level of
artifacts. I was afraid that this op requires more heavy (complicated,
sophisticated- whichever term do you prefere) methods. At my old book
this problem was named as "phase restoration problem". After
diffraction one gets in fact amplitudes of a complex function, so this
function to be restored requires info concerning its phase (to say
nothing about the fact that hi frequence harmonics will be lost during
direct transform). I did only 8 iteration to get my restored images
despite of that in my book similar method requires 600 ones