Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site turtlevax.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!unc!mcnc!decvax!decwrl!turtlevax!ken
From: ken@turtlevax.UUCP (Ken Turkowski)
Newsgroups: net.graphics,net.ai
Subject: Re: Computer Vision, Pattern Recognition --> Generalized Image Files
Message-ID: <833@turtlevax.UUCP>
Date: Wed, 17-Jul-85 18:03:50 EDT
Article-I.D.: turtleva.833
Posted: Wed Jul 17 18:03:50 1985
Date-Received: Fri, 19-Jul-85 01:50:41 EDT
References: <10571@rochester.UUCP>
Reply-To: ken@turtlevax.UUCP (Ken Turkowski)
Organization: CADLINC, Inc. @ Menlo Park, CA
Lines: 46
Keywords: Alpha image component, color spaces
Xref: watmath net.graphics:935 net.ai:2889
Summary: 

In article <10571@rochester.UUCP> sher@rochester.UUCP writes:
>
>Generalized Image Storage Format?
>
>In Ballard & Brown, Computer Vision,  a generalized image is defined
>as an iconic like array containing information relevant to an image
>(paraphrase not quote).  Examples of generalized images are Fourier
>transformed images, edge images, stereo pairs, circle location points,
>image histograms...  If there were generally accepted formats for
>generalized images then I could use edge recognition programs written
>at CMU and image interpretation routines written at U. Mass to test my
>texture recognition routines written here at U. Rochester.  As far as
>I can tell every university stores images differently.  As far as
>other generalized images then every program stores them differently.
>This I believe acts as a gigantic brake on vision research.  

It would be nice if the generalized image could accommodate images used
in computer graphics as well as image processing and machine vision.
One of the more fundamental things about computer graphic images is
that they may not be rectangular, and they may not be opaque (i.e. they
may have some amount of transparency).

Such things are accommodated by including an ALPHA image in addition to
intensity or RGB.  The alpha image acts as a mask, to merge two images,
and can be any number of bits.  One bit alphas are useful as binary
masks which perform a hard switch between one image and another, but
generate jagged edges (and high frequencies!) at the borders between
two images.  The jagged edges resultant from image composition
disappear quickly when using 2 or more bits for alpha.

Additionally, images may be created in different types of color
spaces:  the standard intensity or RGB color spaces are at one end of
the spectrum (pardon the pun) and images painted with a random palette
are at the other.  One type of color space between these two is the one
used by Alvy Ray Smith and Garland Stern for their SoftCel/TintFill
system:  several linearly interpolated color spaces with one common
color at one end of each line.  Image files should contain parameters
that describe (probably symbolically) the color space that the image
was created in.  The image file may also want to include the un-gamma
corrected colormap as well.

-- 

Ken Turkowski @ CADLINC, Menlo Park, CA
UUCP: {amd,decwrl,hplabs,nsc,seismo,spar}!turtlevax!ken
ARPA: turtlevax!ken@DECWRL.ARPA