Path: utzoo!attcan!uunet!husc6!mailrus!tut.cis.ohio-state.edu!bloom-beacon!oberon!neuro.usc.edu!annala From: annala@neuro.usc.edu (A J Annala) Newsgroups: comp.graphics Subject: Re: Palette Optimization Message-ID: <14231@oberon.USC.EDU> Date: 22 Dec 88 10:36:00 GMT References: <12745@cup.portal.com> <571@epicb.UUCP> <9871@gryphon.COM> Sender: news@oberon.USC.EDU Reply-To: annala@neuro.usc.edu (A J Annala) Organization: University of Southern California, Los Angeles, CA Lines: 33 It seems to me that in attempting to develop an algorithm for maximizing the use of a 256 entry color table to represent the spectrum present in a 24+ bit color table image that one might want to us the peak frequency and decay curve of visible light frequencies that can be detected by the three primary (LWS=R, MWS=G, SWS=B cone system) and the sole secondary (rod system) optical image detectors in the human eye. The pigments that are used to capture optical radiation (before transforming it via a high gain chemical [cGMP] amplification cascade) capture light at the the following peak wavelengths: RED cones (long wavelength system) = 419 nm GREEN cones (medium wavelength sys) = 531 nm BLUE cones (short wavelength sys) = 559 nm rods (when simult active w/cones) = 496 nm The ability of the eye to discriminate color differences must be a product of the encoding made possible by light absorbed at these wavelengths. Therefore, a system for compressing the most significant perceived color spectrum in a given image may need to take note of the device characteristics of the primary optical radiation detectors as well as possible post processing by higher cognitive centers. In any case, maybe just one of the computer scientists and/or electrical engineers out there might become interested in using neurophysiological systems analysis of human visual perception to design a more appropriate algorithm for compressing the code for colors represented in a given image. I would be happy to provide references to primary sources for anyone interested in pursuing this approach to the palette optimization problem. AJ Annala, USC Neuroscience Program