FOVEATION SUPPORT AND CURRENT PHOTOGRAMMETRIC SOFTWARE
A. Cöltekin
PL1200, Helsinki University of Technology
02015, Espoo-Finland
Arzu.Coltekin@hut.fi
Commission V, WG V/2
KEY WORDS:
Photogrammetry
ABSTRACT:
Compression, Performance, Virtual Reality, Vision Sciences, Visualization, Foveation, Level of Detail,
Foveation is a term used in computer vision to refer a technique of image compression. This type of compression takes the principals
of human vision into account, trying to mimic the way the "fovea" works. *The human visual system implements an elegant
comprise between the competing goals of maximizing field of view, maximizing spatial resolution, and minimizing neural resources:
[t encodes a large field of view using a retina with variable spatial resolution, and then when necessary, uses high-speed eye
movements to direct the highest-resolution region of the retina (the fovea) at specific points in the visual scene." (Perry & Geisler,
2002)
We can consider foveation as a method of level of detail (LOD) management, or under the bigger umbrella of area of interest
management (AOIM).
Foveation has been employed both when the compression was needed for the system itself and when the image was to be sent over a
network where the bandwidth availability has always been a limiting factor. Consequently, foveation is an efficient technique of
managing photogrammetric visualization, particularly on 1:1 projections or over the network.
This paper presents an introduction of this important technique to the field of photogrammetry.
1. INTRODUCTION
1.1 Motivation
Close range photogrammetry and computer vision techniques
have always had overlapping areas of research and
methodology. This paper discusses one method that is
commonly used in computer vision and image processing but
seemingly not at all in photogrammetry, the method is called
foveation.
Foveation is, in simplest words, a compression method. It is
applied in images and videos that are either presented in big
screens, or transmitted via networks. The first thing that makes
foveation different than other compression methods is its
cognitive quality. It takes the human vision into account.
In the areas where the close range photogrammetry is used, but
particularly when the results are visualized in virtual reality
rooms (caves) and panoramic big screens, this method can be an
effective and suitable way to compress the presented image.
Motivation of this article comes from the realization that the
current photogrammetric software do not have the support for
this method. Therefore it tries to introduce foveation to the field
of photogrammetry while giving an overview of the literature
on the topic.
1.2 Overview
The developments in the computer hardware have been
tremendous over last decades. These developments have
influenced all technical fields, including the 3D graphics and
Virtual Reality (VR) research. Modeling based on microscopic
or telescopic images, optical or thermal, active or passive sensor
data has become possible. At the same time, as visualization
proved useful and fascinating, this created wide interest.
Gaming, scientific visualization, education, documentation of
cultural heritage, architectural walk-troughs, city and terrain
models for planning, navigational or informational purposes
have all blended in with the immersive VR experience.
Even though what is possible today should be viewed with great
respect, it would be fair to accept that it is far from being fully
immersive, and the full extent of what is available is not
available without expensive hardware settings. A more
moderate version of a Virtual Environment (VE) is available on
the Internet offering some interactivity, and that comes with a
set of difficulties.
The amount of data in such environments is big: constructing
detailed vector geometry with polygons and triangles alone can
amount to gigabytes, when rendered graphics and photo-
textures are added, even more. On top of that, if the “virtual
world” includes audio and video, model will be yet bigger.
Internati
Managin
games tl
latest ha
down. Ii
network,
This is ©
out func
transmitt
In this p:
the transi
This app
utilized :
has a cla
things.
idea, he
contingei
“In orde
perceptue
the way i
it is belie
The term
the name
the phen
notes, it i
“The hu
between
maximizi
It encode
spatial re
eye move
retina (th
& Geislei
"Resoluti
than in t
degrees c
held at ar
amount o
of ten. It
the fovea.
our sense
2003)
As can b:
acuity di
towards tl
Figure 2 i
Figure 2
is a 6-fo
or in othe
very
