Copy Link
Add to Bookmark
Report
VISION-LIST Digest 1989 03 08
Vision-List Digest Wed Mar 08 11:00:09 PDT 89
- Send submissions to Vision-List@ADS.COM
- Send requests for list membership to Vision-List-Request@ADS.COM
Today's Topics:
Percent visual input?
OBVIUS (vision software)
VIEWS: Image Processing Toolkit for Suns?
----------------------------------------------------------------------
Date: Tue, 7 Mar 89 11:26 EDT
From: "RCSDY::YOUNG"@gmr.com
Subject: Percent visual input?
Since I am unsure how to reach him via e-mail, I am replying directly to
VisionNet
to Damian Conway's prior question:
> Is the following sentence meaningful?
> "Unimpaired humans receive XX% of all sensory input visually."
Consider there are an estimated 10^9 neurons in the primary visual area (V1)
in cortex, another 10^9 (or possibly much more) in the secondary
visual cortical areas (V2-V4, peristriate, parastriate). Subcortical
areas such as LGN we can disregard, since there are only 10^6 neurons in each
optic tract, and only 10^8 in each eye. In addition we have motion and
eye movement processing over in the superior colliculus and its associated
pathways (perhaps about another 10^9). There are altogether at least
20 different known retinotopic maps in the cortex, not all of which
have complete number estimates. Total cortex is generally thought to contain
about 10^10 neurons, although this figure is a likely underestimate.
The usual estimate of vision-related neurons is made by considering just
the occipital area of the brain, where the visual sensory paths terminate,
which is known to contain about 70 percent of all the neurons in the
human central nervous system (M. D. Levine, Vision in Man and Machine,
1985, p. 84). So 70% would be the most widely quoted figure.
However if you also include association cortex which associates visual
information with auditory and tactile information the total figure would be
higher. Also what about the motor pathways controlling eye movements,
with visual re-afference which is necessary to maintain visual stability
with eye movements? So my own estimate is that final figure would be
that about 80% of the neurons in the brain are involved with
vision processing -- we are indeed visual creatures!
Dick Young
Machine Perception Laboratory
General Motors Research Labs
------------------------------
Date: Tue, 7 Mar 89 17:30:29 EST
From: David Heeger <heeger@paddington.media.mit.edu>
Subject: OBVIUS (vision software)
OBVIUS (Object-Based Vision and Image Understanding System) is an
extension to Common Lisp and CLOS (Common Lisp Object System) for
manipulating pictorially displayable objects. The system provides a
flexible interactive user interface for working with images. In
addition, by using Lisp as its primary language, the system is able to
take advantage of the interpretive lisp environment (the
``listener''), object-oriented programming, and the extensibility
provided by incremental compilation. OBVIUS runs on Sun 3 (using
Lucid Lisp) and Symbolics machines.
The basic functionality of OBVIUS is to present certain lisp objects
to the user pictorially. These objects are refered to as {\bf
viewables}. Some examples of viewables are monochrome images, color
images, one bit images, complex images, image pyramids, image
sequences, filters and discrete functions. A {\bf picture} is a
pictorial representation of a viewable. Note that a given viewable
may be compatible with several different picture types. For example,
a floating point image may be displayed as an eight bit grayscale
picture, as a one bit dithered picture, or as a graphical surface
plot. OBVIUS also provides postscript hardcopy output of pictures.
In the typical mode of interaction, the user types an expression to
the lisp listener and it returns a viewable as a result. The
top-level lisp print function then automatically displays a picture of
the viewable in a window. Each window contains a circular stack of
pictures. Standard stack manipulation operations are provided via
mouse clicks (e.g., cycle, pop, and push). Commonly used operations
such as histogram and zoom are also provided via mouse clicks.
OBVIUS provides a library of image processing routines (e.g., point
operations, image statistics, convolutions, fourier transforms). All
of the operations are defined on all of the viewable types. The
low-level floating point operations on the Suns are implemented in C
for speed. OBVIUS also provides a library of functions for
synthesizing images. In addition, it is straightforward to add new
operations and new viewable and picture types.
OBVIUS is now ready for beta-test distribution (available via
anonymous ftp from whitechapel.media.mit.edu). Since it is currently
an in-house product it comes without warrantee or support. For more
information contact David Heeger (heeger@media-lab.media.mit.edu) of
the MIT Media Lab Vision Science Group, at (617) 253-0611.
------------------------------
Date: Tue, 7 Mar 89 21:26:48 EST
From: achhabra@ucesp1.ece.uc.edu (Atul Chhabra)
Subject: VIEWS: Image Processing Toolkit for Suns?
At a recent conference, I saw a brochure about VIEWS, an image
processing toolkit for SUNs. This is a public domain software
developed at Lawrence Livermore Labs. The brochure contained
the name and phone number of the contact person at Lawrence
Livermore.
I have misplaced the brochure. Could someone on the net email
me the name, phone number, and the email address of the
distributor of VIEWS.
Thanks
Atul Chhabra, Dept. of Electrical & Computer Engineering, ML 030,
University of Cincinnati, Cincinnati, OH 45221-0030.
voice: (513)556-4766 INTERNET: achhabra@ucesp1.ece.uc.edu
OR achhabra@uceng.uc.edu
------------------------------
End of VISION-LIST
********************
