Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!gatech!swrinde!sdd.hp.com!hp-pcd!hpcvsnz!briand
From: briand@cv.hp.com (Brian Dixon)
Subject: Re: HSI to RGB
Sender: news@hpcvsnz.cv.hp.com (News )
Message-ID: <D81238.4KK@hpcvsnz.cv.hp.com>
Date: Thu, 4 May 1995 00:16:19 GMT
References: <3np46d$e66@engr.orst.edu>
Nntp-Posting-Host: hpcvsgen.cv.hp.com
Organization: Hewlett-Packard
X-Newsreader: TIN [version 1.1 PL9.4]
Lines: 38

John Atwood (atwoodj@ada.CS.ORST.EDU) wrote:
: In article <D7nyJy.459@hpcvsnz.cv.hp.com>,
: Brian Dixon <briand@cv.hp.com> wrote:

:  [...]
: >With those diagrams in place, think of HSI-space in the following way:
: >Distance from the central axis on the biconic HSI-space diagram gives
: >the Saturation.  The direction specifies the Hue.

: This helps me understand it, but would'nt the HSI space just be a single 
: upside-down cone?  Under your biconic space, max'ing out both Saturation 
: & Intensity would still taper up the cone to white.  In a single cone model,
: max'ing S & I puts you on the plane of the base of the cone, so that Hue 
: still matters.

: No?

No.  I left out one statement that I thought would be clear from the diagram.
I said that distance from the central axis of the biconic HSI-space diagram
gives Saturation, while the direction around the 'clock' gives the Hue.
What I didn't say that was vertical distance on the central axis (starting
at the bottom cone's point and traversing to the top cone's point) specifies
Intensity.  The top of the top cone is the maximum intensity grey-
scale value, white.  The bottom of the bottom (upside down) cone is the
lowest intensity greyscale value, black.  I know you already thought about
this, so I'm being redundant...but at the maximum intensity value, white,
hue doesn't matter.  It's undefined.  Same goes for the black point on the
bottom cone.  The center point on the plane between the two cones is grey.
Value is 127 on a scale of 0 (black) to 255 (white) for an 8-bit greyscale
system, or 127-127-127 for an 8-bit RGB system.  You can calculate the
HSI values.  In fact, try converting 255-255-255 to HSI and comparing to the
127-127-127 HSI conversion you just did.  I'll do the 0-0-0 case (grinz).

Brian

--
Brian Dixon, Machine Vision Engineer, Hewlett Packard (Corvallis, Oregon)
503-715-3143 (wk), briand@cv.hp.com (email). "Opinions & attitudes are mine!"
