home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.graphics:8223 sci.image.processing:245 comp.ai.neural-nets:2995
- Path: sparky!uunet!munnari.oz.au!comp.vuw.ac.nz!cavebbs!whare!frank
- From: frank@whare.cavebbs.welly.gen.nz (Frank van der Hulst)
- Newsgroups: comp.graphics,sci.image.processing,comp.ai.neural-nets
- Subject: Re: ALGORITHM NEEDED: Palette Selection
- Message-ID: <J12LoB2w164w@whare.cavebbs.welly.gen.nz>
- Date: Tue, 28 Jul 92 20:15:30 NZT
- References: <Bs1wLE.82z@undergrad.math.waterloo.edu>
- Organization: Tech-Soft Systems, Wellington NZ
- Lines: 26
-
- icfaragh@cayley.waterloo.edu (Chad Faragher) writes:
-
- > The situation: An image is stored by giving the rgb value for each pixel.
- >
- > The problem: How to find the best palette of size-n for a particular image
- > stored (at arbitrary resolution) as above.
- >
- > The solution: HELP!!!
-
- .............. is called quantisation.
-
- > Does the algorithm exist? Is it memory intensive? Can it be solved with
- > recursion (ie octree division)? Can a neural-net be trained to arrive at
- > a visually pleasing palette given the original image?
-
- On Simtel PD1:<MSDOS.GRAPHICS> you'll find PVQUAN15.ZIP... that contains
- source code for not one but two quantisers. One does indeed use a
- recursive octree algorithm, the other uses Heckbert's box-splitting
- algorithm. Heckbert's is more memory-intensive (on the PC I used a
- virtual-memory paging system).
-
- Sorry but I can't help you with a neural-net solution... I suspect it
- would difficult to train one to find a "pleasing" solution.
-
- Frank.
-
-