home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.theory.cell-automata:631 sci.fractals:641
- Newsgroups: comp.theory.cell-automata,sci.fractals
- Path: sparky!uunet!enterpoop.mit.edu!galois!riesz!jbaez
- From: jbaez@riesz.mit.edu (John C. Baez)
- Subject: Re: fractal inversion & perpetual motion
- Message-ID: <1993Jan27.025406.23412@galois.mit.edu>
- Sender: news@galois.mit.edu
- Nntp-Posting-Host: riesz
- Organization: MIT Department of Mathematics, Cambridge, MA
- References: <9301220347.AA16147@math.gatech.edu> <POLLACK.93Jan22115242@dendrite.cis.ohio-state.edu>
- Distribution: inet
- Date: Wed, 27 Jan 93 02:54:06 GMT
- Lines: 34
-
- I was interested to read the ad for the Barnsely/Hurd book on
- fractal-based image compression and Jordan Pollack's skeptical reply.
- I too would be fascinated to hear from anyone who has bought the book.
- For those who aren't familiar with the story... Barnsely (and I guess
- Hurd, though I don't know) have greatly hyped an approach to data
- compression for visual images that relies on fitting an iterated
- function system (which produces a fractal) to an image. They have
- apparently claimed compression factors far beyond what normal techniques
- achieve, and demonstrated this to the defense department. Now, one can
- very easily run an iterated function system and get a fractal image.
- I'm sure many of you have seen the very nice picture of a fern one can
- get this way. But the question is, can *most* images be fit well by
- fractals? And more importantly, as Pollack points out, can the fitting
- be done automatically? Barnsely and Hurd seem to claim so... but they
- have never described how they would do so. In fact, Barnsely refused to
- tell Steve Wolfram (of Mathematica fame) how the trick works, and said
- he didn't patent the method, because patenting it would give the secret
- away. This irritated Wolfram, because it implicitly assumed that nobody
- else was smart enough to figure out how this "fractal transform" from
- images to iterated function systems works. Wolfram jokingly said
- (later) that he should figure out the trick and reveal it, just to spite
- Barnesly. I heard about all this (through channels who will remain
- anonymous), and I tried to figure out the trick myself and failed. (I didn't
- try very hard, though, and this sort of thing is not my specialty; I'm
- just a mathematical physicist who knows analysis pretty well.) Now it
- seems that more and more people are thinking there IS no trick - i.e.,
- that the whole thing is a scam.
-
- So - does anyone out there know more about this business? I have no
- desire to make big bucks in the image compression business. It just
- seems to me that if there IS a "fractal transform," it might be a very
- beautiful piece of mathematics, and should not be sitting in some safe
- marked "TOP SECRET".
-
-