International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004
This special feature of human eyes was employed by the
technical researchers as early as 1976. In his paper Clark
ested that objects can be simplified towards the periphery
e field of view. (Clark 1976)
Suge
of th
For the Virtual Environments, this implies the reconstruction of
the visual field according to what is best for human perception.
Because [Virtual Environments are] "synthetic sensory
experiences that communicate physical and abstract
components to a human operator or participant. The synthetic
experience is generated by a computer system that one day may
present an interface to the human sensory system that is
indistinguishable from the real physical world." (Kalawsky,
1993)
Other than satisfying the human senses to provide immersion,
this approach has another function, which is more relevant for
publishing massive graphics and images on the Internet. As
Murphy and Duchowski put it:
“The motivation behind such proposed gaze-contingent systems
is to minimize overall display bandwidth requirements by
reducing peripheral information in concordance with the
perceptual limits of the Human Visual System (HVS). Efforts at
providing peripherally degraded information date back to early
eyeslaved flight simulators.” (Murphy & Duchowski, 2001)
3. TERMINOLOGY IN ACTIVE VISUALIZATION
As the reader might have noticed, the last quote did not mention
"foveation" but a system that was “gaze-contingent”. As human
vision is explored by different research groups, this concept is
also called with different terms. The terms carry slightly
different associations; for instance foveation, as a term, is used
for systems with or without the eye-tracking while gaze-
contingent term is associated with an eye-tracking system. They
essentially mean the same thing.
For giving a name to this phenomenon, other than “foveated”
and “gaze-contingent”, also “Eccentricity LOD” is used. Reddy
uses this last term in his PhD thesis on “Perceptually Modulated
LOD”. (Reddy 1997)
In his thesis, Reddy summarizes the techniques used for
controlling level of detail in VR, where 5 implementations of
LOD are listed:
- Distance LOD
=. Size LOD
- Eccentricity LOD
- Velocity LOD
- Fixed Frame Rate LOD
Area of Interest Management (AOIM) is a more containing
word, which covers all means of LOD, including the
Eccentricity LOD (i.e. foveation). In 2D image coding, this
term corresponds to Region of Interest (ROI) and in eye-
tracking systems Point of Regard (POR) has a similar hold.
In AOIM, by including the word “management”, the
implication is moved from the area or the region, to the fact that
something is done to manage it.
To summarize the umbrella of the terminology, the following
bulleted list should also express relationship between these
terms:
- Active Visualization
- Area of Interest Management (AOIM) (an active visualization
method)
- Level Of Detail (LOD) (an AOIM method)
Terms used for human vision based LOD approaches:
- Gaze directed
- Gaze Contingent
- Perceptually Driven
- Eccentricity
- Foveated
Level of Detail management includes foveation but deals with
other directions and ways the perception of detail varies as
mentioned in the previous page. AOIM refers to the approaches
that include LOD, while ROI and POR are used similarly as
AOIM. All of these concepts are met when dealing with Active
Visualization.
Note that, the term *perceptually-driven" in the literature covers
a little more than foveal approach, it generally refers to a more
generic account for what we visually percept, which, then
includes distance-LOD and velocity-LOD and such. In many
cases though, the term does not cover other senses — therefore it
is not a complete reference to human perception.
“Active Visualization” is used similarly as Area of Interest
Management (AOIM), although it implies a broader coverage.
There are a number of other relevant and similar terms such as
smart sensing, visibility determination, attentional scene
segmentation, active perception — which are used to refer to the
active visualization.
4. COMMON FOVEATION METHODS AND A FEW
EXAMPLES
Can foveation be useful without eye tracking?
Before continuing with the methods and examples in the last
chapter of this paper, it would be good to spare a few lines to
explain why we are not concentrating on systems with eye-
tracking equipment.
The foveation concept, by its description, is often closely
related to tracking the cyes, or at least the head position of the
viewer. The view-dependent real time rendering of the scene
would require knowing the view point at all times, therefore it
is obvious that one would want to have the eves of the user
tracked instantaneously.
But eye-tracking systems are complex and still very expensive.
It is expressed that the expense is not because of the cost of the
hardware, but because of the small market (UTEXAS 2002).
Same source claims that when VR becomes more widely used,
it would essentially include the tracking systems with it.
Until the cave VR experience is widely available to public, the
notion of it can be passed via network-based interactive 3D
environments, using the “worlds” that are created with VRML
(Virtual Reality Modeling Language), Java3D and similar.
There is currently no eye-tracking in experiencing the online
interactive 3D models, nor they are fully immersive, but they
certainly are available to a much larger audience. Along with
their many other useful functions, they serve as a first
introduction to VR and give an understanding of its potential.
Internatic
In such s
compared
to test th
for now, \
It is possi
of an av
equipmen
of system
as “fishte
accessork
yet. Whei
ofa VE t
assumed.
When res
cannot be
tracking t
known at
looking a
interactive
with an o
there. Thi
method
implemen
For the F
tracking t
at the poit
intermedi:
will be m:
2D Fovea
Foveation
image CO
point, a bi
the appro
large nun
technique:
The comr
pass filte
decoding
make it i
2002, Ch:
1996, Red
Use of G
dominate
along witl
implemen
(GPL) by
"Foveator
in (UTEX
and the so
