The Challenge of
Visualizing Three-dimensional Data
VRVis Research Center in Vienna, Austria
Computer graphics has become an integral part of everyday life and
work. As one part of computer graphics, visualization
plays an important role when data from various sources
should be presented or investigated. One example is the
visualization of 3D data, which originates in computer
tomography or similar acquisition modalities. Here, the
special challenge of visualization is to make meaningful
images of data which densely populates the 3D domain. The
problem to deal with is occlusion, of course. 2D images
just do not provide sufficient space for all data to
be visualized. Quite some solutions have been presented
up to now to nevertheless generate useful images. Direct
volume rendering on the basis of compositing,
maximum-intensity projection, rendering of
boundary-structures like iso-surfaces, etc., are just
the most important techniques to be mentioned here. This
talk tries to demonstrate the challenge of 3D
visualization by giving an intuitive visualization of
the problems first, as well as by showing solutions,
which have been established over the last years.
One reason, why visualization has become very popular
during the past decades, is that human observers usually
make effective use of their powerful visual system when
investigating their environment. Especially the analysis
of images, which have been perceived through the human
visual system, is very effective. Features are quickly
detected, even without the need for serial search .
And that is exactly the point where visualization hooks
in - instead of listing large amounts of numbers as a
direct representation of data visual depictions are used
to effectively convey information. However, the bandwidth
of the human visual system also is limited, and therefore
the amount of information which can concurrently be provided
to the human observer is limited also.
The bandwidth of the human perception can be measured in
different terms such as resolution, extension, and dimension.
The latter will be of most impact for the following considerations.
From the spatial placement and orientation of both human eyes,
we can derive that human vision only is of a little more than 2D.
All human 3D vision amounts to the perception of planar data
which is only enhanced by depth information. Human observers
usually see boundary surfaces of solid 3D objects which surround
them. From internal and automatic correlation analysis between
the images of both eyes as well as from shading of surfaces the
3D shape of such surfaces is derived.
The challenge of visualizing 3D data now is to effectively
communicate 3D information as such which is provided by
3D scanning modalities in medical applications
while at the same time dealing with the limited bandwidth
of the human visual system. As the above described limitation
does not allow to directly present 3D data to visualization
users - which in this case would degrade visualization to a
trivial job - indirect methods are required to support the
investigation of 3D data. Either subsets of the data are
shown, or aggregations thereof. Also the use of indirect
representation through boundary surfaces is often used.
Many approaches have been presented in the past which were
tailored to effectively convey 3D information, more specifically
to gain useful insight about the interior of 3D objects.
In this talk a few of them are discussed together with their
advantages and disadvantages. Surface-based methods as, for
example, those which are based on
the Marching Cubes algorithm  are compared to
direct volume rendering (DVR)
as initiated by Kajiya and Levoy [6,8].
For DVR, the special problem of how to intuitively specify a transfer
function is presented together with a couple of
The problem of stacking iso-surfaces is addressed as discussed
by Interrante  as well as focus-plus-context
approaches for volume rendering are discussed as, for example,
two-level volume rendering . Also, non-photorealistic
rendering of volumetric data as described
by Ebert and Rheingans  is compared to other approaches
such as maximum-intensity projection 
or simple data-integration.
Finally, this talk will conclude with the observation that the choice
of the most appropriate visualization technique - as there is no best
available which would allow to trivially show all the data -
heavily depends on the data given, its internal structure,
as well as the goal of the visualization user.
The Challenge of
Visualizing Three-dimensional Data
C. L. Bajaj, V. Pascucci, and D. R. Schikore.
The contour spectrum.
In Proc. of IEEE Visualization '97, pages 167-174.
D. Ebert and P. Rheingans.
Volume illustration: non-photographic rendering of volume models.
In Proc. of IEEE Visualization 2000, pages 195-202.
H. Hauser, L. Mroz, G.-I. Bischi, and M. E. Gröller.
Two-level volume rendering - fusing MIP and DVR.
In Proc. of IEEE Visualization 2000, pages 211-218.
T. He, L. Hong, A. Kaufman, and H. Pfister.
Generation of transfer functions with stochastic search techniques.
In Proc. of IEEE Visualization '96, pages 227-234.
V. Interrante, H. Fuchs, and St. Pizer.
Illustrating transparent surfaces with curvature-directed strokes.
In R. Yagel and G. Nielson, editors, Proc. of IEEE
Visualization '96, pages 211-218.
J. T. Kajiya.
Ray tracing volume densities.
In Proc. of ACM SIGGRAPH '84, pages 165-174.
G. Kindlmann and J. Durkin.
Semi-automatic generation of transfer functions for direct volume
In Proc. of IEEE Volume Visualization '98, pages 79-86.
Display of surfaces from volume data.
IEEE Computer Graphics & Applications, 8(5):29-37, 1988.
W. E. Lorensen and H. E. Cline.
Marching cubes: a high resolution 3D surface construction
In Proc. of ACM SIGGRAPH '87, pages 163-189.
J. Marks, B. Andalman, P. A. Beardsley, W. Freeman, S. Gibson, J. Hodgins,
T. Kang, B. Mirtich, H. Pfister, W. Ruml, K. Ryall, J. Seims, and S. Shieber.
Design galleries: A general approach to setting parameters for
computer graphics and animation.
In Proc. of ACM SIGGRAPH '97, pages 389-400.
Preattentive processing in vision.
Computer Vision, Graphics, and Image Processing, 31:156-177,
K. J. Zuiderveld, A. H. J. Koning, and M. A. Viergever.
Techniques for speeding up high-quality perspective maximum intensity
Pattern Recognition Letters, 15:507-517, 1994.
This document was generated using the
LaTeX2HTML translator Version 99.1 release (March 30, 1999)
Copyright © 1993, 1994, 1995, 1996,
Computer Based Learning Unit, University of Leeds.
Copyright © 1997, 1998, 1999,
Mathematics Department, Macquarie University, Sydney.
The command line arguments were:
latex2html -split 2 invTalk
The translation was initiated by Helwig Hauser on 2001-04-18