The Third International Conference in Central Europe on Computer Graphics and Visualization 1995

University of west Bohemia, Pilsen, Czech Republic

February 14 - 18, 1995

Abstracts of papers

 
******************** V O L U M E   I **************************             

GENERATING A MODEL OF PLANT USING NURBS

        Bedrich Benes 

CTU, Fac.of Electrical Eng., Dept. of Computer Science
Karlovo nam.13, 131 25 Prague, Czech Rep.
e-mail benes@sgi.felk.cvut.cz

Abstract is not available.

Keywords: Computer Graphics, NURBS, Modelling


ONE METHOD OF THE HUMAN BODY MODEL ANIMATION

            R. Berka, I. Jelinek 

Department of Computers
Faculty of Electrical Engineering
Czech Technical University
Karlovo nam. 13
121 35 Prague
Czech Republic            
e-mail: berka@cs.felk.cvut.cz
        jelinek@cslab.felk.cvut.cz 
 
In the first part of the paper we show a new approach to the 
modelling of the human body. The modelling of the human body 
has two parts - the model structure description (the model of 
the human body we call "figure") and the model motion control.
The second part of this article is devoted to methods of the
figure's structure formal description and to control of the 
figure's behavior. For this purpose we have proposed a 
special formal language FDL (Figure Description Language). 
The next part informs on the implementation our ideas in 
program VITALIATOR. The implementation is consistently based 
on the object oriented approach. In the last part of our 
contribution we discuss questions connected with control of 
all actor's body parts and parallel movement control of more 
figures.

Keywords : computer animation, object-oriented graphics, 
human body animation, modelling


AN APPROACH TO DRAWING FAIR PLANE OPEN CURVES

   Marcelo Cordeiro Bernardes, Wu Shin-Ting, 
             Itala M.L.D'Ottaviano 
             
DCA - FEE - UNICAMP, CLE/IFCH - UNICAMP             
e-mail: {ting,mcb}@dca.fee.unicamp.br
         itala@cle.unicamp.br

In this paper an approach for interpolating a given sequence 
of points by a fair plane curve is presented. Since the 
fairness concept is subjective, a non classical modelling 
tool - fuzzy sets - is used and interactive facillities are 
provided. Some results of our implementation are included.


A PARALLEL APPROACH TO HIERARCHICAL RADIOSITY

      Christian A. Bohn, Robert Garmann 
      
German National Research Center for Computer Science      
Dept. Visualization and Media Systems design
Sankt Augustin, Germany
e-mail : bohn@viswiz.gmd.de, 
         garmann@ls7.informatik.uni-dortmund.de

A parallel algorithm solving the radiosity equation is 
presented. It is based on the hierarchical approach (HR) [5] 
and realized on a massively parallel supercomputer - the 
Connection Machine 5. 
	Our algorithm considers the HR approach as a process 
that manipulates a huge graph structure. Simulated annealing 
is used in the graph's rearranging procedure to achieve 
a good work-ballance and nearly optimal communication costs. 
	The implementation shows a significant step to 
faciliate the aplication of a radiosity solver, produced on 
one hand by the few user-support that HR needs, on the other 
hand by the fast calculation times the parallel 
implementation offers. On 64 processors we obtained a speed-up
of 8.4.

Keywords : Computer Graphics, Scientific Visualisation, 
Hierarchical radiosity, Rendering, Simulation, 
Parallelization, Message Passing, Supercomputer, 
Architecture.


FAST COLOUR SHADING

    J.J.Bourdin 

Departement d'Informatique
Universite Paris-8
2,rue de la Liberte
935 26 Saint-Denis 02 - France
e-mail: jj@aristote.univ-paris8.fr

Colour shading a region of the plane is the action of filling 
its interior with continuously varying colours. Some classic 
methods are presented. This paper proposes an original 
approach to colour shading. A condition of smoothness is 
given. A region is modeled as a support and a chromatic 
function. Various shapes of shading are presented. 
Incremental computing and benchmarks are also discussed. In 
addition, we solve the problem of good visual smoothing for 
an angular colour shading.

Keywords : Shading, colour shading, 2D Graphics, Computer 
graphics, fundamental algorithms


STUDY OF ITERATIVE PROCESSES IN COMPUTER GRAPHIC'S PROBLEMS

Jean-Marc Brun, Sebti Foufou, Abdelaziz Bouraz, 
Philippe Arthaud 

LISPI /LIGIA bat.710
Universite Claude Bernard LYON 1
43 Bd du 11 Novembre 1918
69 622 Villeurbanne Cedex

Coretech International
27-29, Bd du 11 Novembre 1918 B.P. 2131
69 622 Villeurbanne Cedex

The convergency of iterative processes can exhibit unexpected
behaviour. In this paper an analysis of the convergency of 
the iterative processes is made through the aplication of the 
Newton method to the polynomial equations resolution. This 
leads to the definition of a dynamic relaxation scheme 
applied to computer graphics.

Keywords : Iterative processes, convergency, relaxation, 
equations, graphics.


THE INTERACTION OF THEORY AND PROCEDURE IN FRACTAL GEOMETRY

                     Stephan D. Casey 
                     
Department of Mathematics and Statistics
The American University
4400 Masachusetts Ave., N.W.
Washington, D.C. 20016-8050                     
e-mail:scasey@american.edu

This article discusses the interplay in fractal geometry 
occuring between computer programs for developing 
(approximation of) fractal sets and the underlying dimension 
theory. The computer is ideally suited to implement the 
recursive algorithms needed to create these sets, thus giving
us a laboratory for studying fractals and their corresponding 
dimensions. Moreover, this interaction between theory and 
procedure goes both ways. Dimension theory can be used to 
classify and understand fractal sets. This allows us, given 
a fixed generating pattern, to describe the resultant images 
produced by various programs. We will also tie these two 
perspoectives in with the history of the subject. Three 
examples of fractal sets developed around the turn of the 
century are introduced and studied from both classical and 
modern viewpoints. Then, definitions and sample calculations 
of fractal and Hausdorff-Besicovitch dimensions are given. 
Finally, dimension theory is used to clasify images.


TEXTURE, DISPLACEMENT AND IMMERSION : A MODEL FOR TREE 
                       RENDERING
                       
Miguel Chover, Roberto Vivo, Ricardo Quiros, Xavier LLuch 

Departamento de Informatica, Universidad Jaume I.
Campus de Penyeta Roja, 12071 Castellon, Espana
Tel. ++34 64 345771, fax ++34 64 344858
e-mail chover@inf.uji.es

Departamento de Sistemas Informaticos y Computacion
Universidad Politecnica. Camino de Vera s/n
46071 Valencia, Espana
 
This paper presents a new method to generate realistic 
details on tree surfaces and structures. The method combines 
texture, displacement and immersion techniques. We use 
an extension of L-systems, namely random L-Systems, to 
construct the tree, and represent its structure of branches 
using parametric surfaces (cones). Adding displacement at the
surface level by means of 2D and 3D functions, and deviation
at the structure level by immersion in a 3D function, we 
perturb the surface definition. The parametric definition of
the surface is adequate to apply textures and reduces the 
spatial cost of the previous tree representations. In 
addition, the proposed technique can even manage further 
detail than previous ones.


DYNAMIC SUBDIVISION IN RADIOSITY

  Eric Dubuis, Hanspeter Bieri 

Institut fur Informatik und angewandte Mathematik
Universitat Bern, Langgasstrasse 51, 3012 Bern, Switzerland
e-mail dubuis@iam.unibe.ch, bieri@iam.unibe.ch

This paper presents a fast radiosity algorithm for 
illuminating scenes containig large piecewise polygonal 
surfaces. Dynamic subdivision is based on the well known 
Adaptive Subdivision, introduced by M.Cohen et.al.[3]. During
the illuminating process, patches in areas  with high 
intensity gradients are refined. Contrary to Adaptive 
Subdivision, the presented algorithm subdivides patches not 
in a static way. The patch hierarchy is dynamic and adapted 
to the respective state of the illumination process. To take 
a decision concerning path subdivision, more information 
about the gathered energy of a patch is considered than with 
Adaptive Subdivision. The results show that this new 
algorithm can lead to remarkable speedups compared to 
Adaptive Subdivision.


SPEEDING UP PROGRESIVE RADIOSITY BY OVERSHOOTING

               Martin Feda 
               
Technical University of Vienna
Institute of Computer Graphics
Karlsplatz 13/186-2
Vienna, Austria               
e-mail feda@cg.tuwien.ac.at

Overshooting techniques have proven to significantly speed up
the convergence of radiosity computation. Similar methods 
(overrelaxation) have also been widely used in numerical 
mathematics to speed up iterative solution methods for linear
equation systems, e.g. Gauss-Seidel iteration. This paper 
gives a comprehensive description of all overshooting 
techniques. Advantages and disadvantages of the different 
techniques are also described.


ATTRACT - ITERATIVE VISUALIZATION OF DYNAMICAL SYSTEMS

          Eduard Gröller, Herbert Oppolzer 
          
Technical University Vienna
Institute of Computergraphics
Karlsplatz 13/186/2
A-1040 Vienna, Austria
Tel.+43(1)58801-4582
FAX +43(1)5874932          
e-mail groeller@cg.tuwien.ac.at
        herbert@cybercafe.co.at
        
An interactive program for visualizing the long term behavior
of dynamical systems, e.g., attractors and bifurcation 
diagrams is presented. The program allows an easy 
specifications of a set of formulas and constants which 
describe a dynamical system. This set of equations is then
used for displaying the geometrical  structure of the long 
term development of dynamical system. The user defines the 
assignment of variables of the dynamical system to at most 
three spatial axes and one colour axis. Viewing parameters,
e.g., point of view, zoom, rotation angle, can again be 
changed interactivelly. The program is intended to provide 
researchers working on dynamical systems with a fast visual 
analusis and experimentation tool.

Keywords : visualization, dynamical system, interactivity, 
attractor, bifurcation diagram.


COMBINING CLUSTER ANALYSIS AND CONTINUOUS ISOSURFACE 
       COLOURING IN A TOOL FOR DATA EXPLORATION
       
                     Johan Hangman 
                     
SSKKII, University of Göteborg
41298 Göteborg, Sweden
e-mail hagman@ling.gu.se

The paper shows how two powerfull techniques for supporting 
data exploration of multidimensional data can be combined in 
a tool for this purpose. The techniques, cluster analysis and 
graphics visualization, are briefly presented and discussed 
as modules of a prototype. Its preformance is illustrated by 
experiments resulting in cluster configurations where the 
value distribution of "underlying" dimensions are visualized 
with colors or shades of grey.


NON-PHOTOREALISTIC SHAPE CUES FOR VISUALIZATION

                     Peter Hall 

Department of Computer Science
PO Box 600
Victoria University of Wellington
New Zealand                     
e-mail peter@comp.vuw.ac.nz

Visualization applications often need to display objects that
are both three dimensional and coloured. The objects may be 
represented either by surfaces or by volumes. The motivation 
behind such displays is to demonstrate two or more 
independent variables at once. So one variable may determine 
object shape and anotehr variable may determine the 
distribution of colour. A lighting model is needed to provide
cues for visual shape recognition. Conventional lighting 
models provide these cues by altering colour. Hence 
a conflict arises when this approach is adopted because the 
total color variation is made up of two components that are 
perceptually indistinguishable. There are variations that 
show the distribution of a variable, and global variations 
caused by the lighting model. It might be said that a side 
effect of the light model is to distort the information 
represented by colour. 
	This paper offers a solution to this problem be 
replacing the conventional lighting model with a structured 
lighting model of the kind used in machine vision context. 
The proposed solution simulates the shadow of a regular, 
planar grid on the surface of an object. Improved shape cues 
can be obtained by using a grid that is sensitive to lighting 
conditions, this generates a cross-hatch shading effect. The 
notion of a planar grid easily generalises to three 
dimensions, where it can be used not only with surfaces, but 
also with volumes. The benefit of the method is that colour 
is effected only locally, and in well defined places; the 
remaining colour is readily identifable and faithfully 
represents underlying data. In addition, the method in either 
of its forms is simple to incorporate into existing 
renderers.
	The method is demonstrated with results obtained from 
a computation fluid dynamics application. Both coloured 
iso-surfaces and coloured volumes are shown.


ADVANCED DESIGN SOFTWARE FOR THE INJECTION MOULDS DESIGN 
          BASED ON SOLID MODELLING TECHNIQUES
          
                     Ir. T. S. Huang 

Division of Production engineering, 
Machine design and Automation                     
Faculty of Engineering, Department of Mechanical Engineering
Katholieke Universiteit Leuven
Celestijnenlaan 300B
B-3001 Heverlee Belgium

This paper describes a CAD software that focuces on 
computer-aided design for injection moulds using solid 
modelling techniques. The ultimate goal of this CAD software 
is to enable the designer to do various mould design works. 
It consists of two main modules, geoetric modelling and mould
design. The former consists of a set of functions which can 
be used to create solid primitives and execute Boolean 
operations and the latter is able to support the design of 
standard mould components and some further applications such 
as the automatic determination of undercut, parting lines and
parting surfaces, core and cavity plates, etc. This software
is now implemental on a Unigraphics CAD/CAM system and 
operated in a dialogue mode to communicate with the user.

Keywords : Solid modelling, three-dimensional convex hull, 
Injection moulds, Undercuts, Parting lines and parting 
surfaces.


OPTIMIZING PATH TRACKING USING NOISE REDUCTION FILTERS

      Henrik Wann Jensen, Niels Joergen Christensen 
      
Dept. of Graphical Communication
Technical University of Denmark
fax : +45 45 93 83 17 phone : + 45 45 93 41 66      
e-mail : igkhwj@unidhp.uni-c.dk, igkncj@unidph.uni-c.dk

The problem of global illumination can be solved using path 
tracing. Unfortunatelly path tracing gives very noisy images.
This noise is mostly caused by the indirect illumination 
reflected diffusely. The common way to reduce the noise is 
to use more samples/rays per pixel. However, tho convergence 
speed in path tracing makes this strategy very costly. In 
this paper we propose a technique in which light reflected 
diffusely two times is separated from the final solutin. This 
light is filtered by different noise reduction filters and 
then added to the remaining solution. In this way we avoid 
blurring the image and at the same time we are able to reduce 
the noise level significantly.

Keywords : Global illumination, path tracing, noise reduction,
filtering.


GRAPHSUPPORT : INTERACTIVE MODELLING WITH COMPUTER GRAPHICS 
                    IN ASSISTING DESIGN
                    
                      Ö. Karaçali 
                      
University of Technology, Loughborough, England                      
e-mail O.karacali@lut.ac.uk

This paper explains a novel approach that the dsesigner can 
be supported by interactive object modifications with 
graphics system. An engineering application, called 
GraphSupport system is identified and relations to the 
software support aspect is discussed. An example of 
industrial product in interactive modelling with objects is 
applied to GraphSupport to aid engineering design. Simple, 
yet versatile, information and functional modelling methods 
are introduced , allowing variants and views of the product
to be represented and modified in a solid modelling 
environment.


EFFICIENT FILTER COMPUTATIONS WITH SYMMETRIC MATRIX KERNELS

                       Manfred Kopp 
                       
Institute of Computer Graphics, Visualization and Animation Group
Technical University of Vienna
Karlsplatz 13/186-2, A-1040 Wien
e-mail m.kopp@ieee.org, http://www.cg.tuwien.ac.at/~kopp/

This paper presents an algorithm for filter calculations using 
symmetric matrix kernels. This algorithm outperforms 
traditional methods for kernels larger than or equal to 5x5 
on machines based on RISC designs, where the time needed to 
calculate an adittion equals the time needed for 
multiplication. The algorithm is based on a decomposition of 
the kernel matrix into several kernel matrices of decreasing 
size, which can be computed very fast because of
spatial coherence. A comparison with traditional methods 
shows the efficiency of the presented approach.

Keywords : image processing, anti-aliasing, filtering, 
symmetric matrix kernels, spatial coherence.


PARAMETRIZING SUPERQUADRICS

Helwig Löffelmann, Eduard Gröller 

Institute of Computer Graphics
Technical University of Vienna
A-1040 Karlsplatz 13/186/2

Superquadrics are well known and often used 3D surface 
objects in computer graphics. They are used for modelling 
parts of scenes that are then rendered using photorealistic 
synthesis algorithms (e.g., ray tracing) . For some 
techniques, like texturing, which are part of these rendering 
methods, the type of the parametrization of such a surface 
has to be chosen carefully and is not intuitively obvious at
first sight. There are cases, where the straightforward 
extension of quadric parametrizations to superquadrics do not 
produce satisfying results. We therefore investigate a number 
of different parametrizations in combination with the 
corresponding formulas, and point out some significant 
differences between them.

Keywords : Conics, Superconics, Quadrics, Superquadrics,
Parametrizations


TOPOLOGICAL OPERATIONS FOR NON-MANIFOLD MODELING

        David Marcheix, Stefka Gueorguieva 

Processing non-manifold topologies (abbreviated as NMT) plays 
a primary role in current development of solid modelling 
applications. Euler operators are a powerfull tool for 
creation of valid boundary representations of solids. By 
contrast with the manifold domain, where topological 
operators are well understood and implemented, there is a 
lack of elaboration of their non-manifold counterparts. 
	In this paper we work out a basis of Euler-like 
operators for construction, maintenance and manipulation of
boundary schemes of non-manifold objects. The presented 
algorithms are implemented for a NMT data structure and 
experimental results are discussed in the context of the 
completeness and soudness of the NMT operators.

Keywords : geometric modelling, non-manifold topology, 
boundary representation, Euler operators, data structures, 
algorithms.


AN AVERAGE-CASE COMPLEXITY OF RAY TRACING ALGORITHMS

         Gabor Marton, Laszlo Szirmay-Kalos 
         
Department of Process Control
Technical University of Budapest
Budapest, Müegyetem rkp.9/R, H-1111, Hungary         
e-mail marton@seeger.fsz.bme.hu

A theoretical framework for analyzing average-case time and 
storage complexity of ray tracing acceleration techniques 
is introduced by means of homogeneous spatial Poisson point 
processes. Then, as a demonstrative example of its 
application, the expected query time of the widely known 
techique based on a regular spatial grid is analyzed. Finally,
an interpretation of the results is presented within the 
context of probability theory.


USER-COMPUTER INTERACTION : COGNITIVE PROPERTIES OF ICONS 
            FOR MULTIDIMENSIONAL DATA ANALYSIS

Antonio Messina, Laura Moltedo, Silvana Contento, 
                 Roberto Nicoletti 

Dipartimento di Fisica "A.Righi", Universita di Bologna
Instituto per le Applicazioni del Calcolo "M.Picone" , Roma
Dipartimento di Psicologia, Universita di Bologna
Dipartimento di Scienze Biomediche, Universita di Modena
Italy

In the graphical environment, icons (i.e. symbols or 
pictorial representations analogous to physical objects, 
actions or functions) are generally more informative and 
easier to manipulate than verbal labels. Because icon should 
resemble the functions they represent, pre-testing should be 
done to determine whether the user understands what the icons 
are supposed to symbolize. In the present work we aimed at 
establishing whether the criteria adopted to shape the icons 
of a graphic tool for the analysis of multidimensional field 
(MUDI3) were shared by the user. Method, procedure and 
results are presented and discussed.


ON SCENE COMPLEXITY DEFINITION FOR RENDERING

Ludovit Niepel, Jozef Martinka, Andrej Ferko, Pavol Eliass 

Department of Computer Graphics and Image processing
Comenius University, 842 15 Bratislava, Slovakia
e-mail niepel, ferko, elias@fmph.uniba.sk
Jozef.Martinka@st.fmph.uniba.sk

This paper introduces one way how to define mathematically 
correct way the complexity of the polygonal scene for 
rendering. This complexity allows to compare previously 
uncomparable rendering algorithms via computing the 
complexity of various scenes, and give the more rigid 
solution of other related problems. The advantages of this 
unified approach are discussed.


PARALLEL ADAPTIVE RAY_TRACING

 Irena Notkin, Craig Gotsman 
 
Department of Computer Science
Technion - Israel Institute of Technology
Haifa 32000, Israel
e-mail {cyrus|gotsman}@cs.technion.ac.il

We describe a dynamic task allocation algorithm for ray 
tracing by adaptive progressive refinement on a parallel 
computer. Parallelization of adaptive ray tracing is 
difficult because of the inherent sequential nature of 
the sample location generation process, which is optimized 
(and different) for any given scene. We report on 
experimental results obtained from our implementation of 
this algorithm on a Meiko parallel computer. The three
performance measures of the algorithm, namely, load-balance,
speedup, and image quality, are shown to be good.


TUNNING RENDERED LINE-DRAWINGS

Bernhard Preim, Thomas Strothotte 
e-mail {bernhard|tstr}@isg,cs.uni-magdeburg.de

Normally rendered imagesare the polished output of rendering 
software and are, by definition, "perfect". However, as we 
think about rendering abstractions of the underlying models, 
rather than photorealistic images, it becomes evident that 
user involvement is necessary to tune the rendered images to 
his or her satisfaction.
	In this paper we outline methods to tune rendered 
images. We concentrate on-line drawings, as these are most 
readily edited to make the modelled object appear more 
abstract. Rather than having user edit individual lines, our
emphasis lies on studying, what users really want to achieve, 
and how we can support them more directly to attain these
effects. This emphasis is used to guide the design of the 
user-interface to the system for tuning line-drawings.
        We present a tool which allows high-level 
intent-based interaction to generate line-drawings from 
rendered 3D-models. Included in the system are editting 
facilities to fine-tune and thereby individualizing 
line-drawings. High-level interaction draws on statistical 
methods to distribute over the rendered image. We show how 
the interaction can be improved if information about the 
model is available to the image editor.

Keywords : Non-photorealistic images, line-drawings, 
architectual sketches, human-computer interaction.


PHOTO-REALISTIC GRAPHICS FOR EMERGENCY LIGHTING VISUALISATION


              Thomas Ramstad, Alan Chalmers 
              
Department of Computer Science
University of Bristol
Bristol, BS8 1TR
United Kingdom

In an emergency situation it is vital that people are rapidly 
able to identify the exits. This is particularly important 
when the visibility of the sign can be significantly reduced, 
for example in a smoke filled room. There are several factors
which contribute to a sign's visibility : the colour of the 
sign, luminance, size of letters, the location of the sign in 
the room and of course how engulfed the room is with smoke. 
It is also important to determine the visibility in clear and 
dark conditions. This paper describes a visualisation system 
under development which uses photo-realistic computer 
graphics techniques and parallel processing to provide an 
interactive tool for emergency lighting designers. The system
will enable designers to investigate different positioning 
and illumination of emergency lighting under a variety of 
environmental conditions.


VISUALIZATION OF SMOOTHED PARTICLE HYDRODYNAMICS

                  Rene T. Rau 
                  
WSI/GRIS, Universität Tübingen
Auf der Morgenstelle 10
D-72076 Tübingen, Germany                  

Today smoothed particle hydrodynamics (SPH) play an important 
role in the computation of gas dynamical processes appearing 
in astrophysics and other fields. Nevertheless the 
visualisation techniques for this particle based model are 
still rudimentary. In order to visualize the density and 
pressure appearing in the simulation correctly we 
investigated the simulation in more detail. By this 
investigation we show that a splatting based visualization 
technique displays the volume density and pressure used in 
the simulation exactly. Furthermore we conclude that in our 
situation splatting is the only reasonable technique and is 
superior to other techniques such as raycasting. Finally, our
visualization method enables the scientist to control the 
result of the simulation especially with few particles and 
gives high quality rendering at every state of the research 
process.


******************** V O L U M E   II **************************             

 
MULTIMEDIA INTERFACE FOR ADVANCED MATHEMATICS:
          OTTER 3.0 AS A CASE STUDY

              John F. Richardson 
              
NCCOSC
Code 44201
53560 Hull Street
San Diego, Ca. 92152
United States              
e-mail richards@marlin. nosc.mil

This paper describes a multimedia interface for the OTTER 
theorem proving system. This interface  is implemented using 
an advanced multimedia authoring system based upon the 
SUPERCARD program. SUPERCARD is a Hypercard type system that 
basically solves most of the problems prevoiusly existing in 
Hypercard. This interface described in this paper is designed 
for the Macintosh computer although multimedia authoring 
systems exist for Unix and PC based systems.
	The primary purpose of this interface is to provide 
a "Macintosh look and feel" environment for the OTTER theorem 
proving system. The interface provides this look and feel 
through the use of hypermedia objects called cards. These 
cards serve as backdrops for interactive graphics and 
controls made available to the user for interacting with the 
theorem power.
	The secondary purpose  of the interface is  to 
provide a natural environment for navigating within the 
theoretical realms of Automated Reasoning, Logic Programming,
Theorem proving and mathematics. The interface seeks to 
restructure the rules and symbols of the above realms into 
more "user friendly" and "intuitive" rules and symbols for 
use by students and general mathematicians. The interface 
also seeks to make it easier on experts in the field of 
Automatic theorem proving.

Keywords : Multimedia, Automatic Theorem Proving, 
Expert Systems, Animation, Hypermedia


FLOW VISUALIZER

 Harald Ruckser 

Institut für Informatik
Abteilung für graphishe und parallele Datenverarbeitung
Universität Linz, Altenbergerstrasse 69, 4040 Linz
Austria

In this paper a new system for visualization of combined 
scalar and vector fields is presented. The methods 
implemented in this system are based on the use of 
combination AVOs (abstract visualization objects)  consisting 
of arrows, spheres, isosurfaces, streamlines and cylinders. 
These AVOs improve the understanding of correlations between 
the spatial distributions of one vector and up to three 
scalar quantities, and enable simple forms of topological 
filtering. Furthermore, new clasification methods for 
visualization methods and AVOs are introduced, which, among
others, describe the dimensionality of the data that a single
AVO displays (DD - displayed dimensions). An extended model 
of the visualization process reveals important details of the 
microstructure of visualization, its transformational phases 
and objects.


AN ARCHITECTURE OF MULTIMEDIA PLATFORM FOR INTEGRATED 
            AUDIO-VISUAL DATA PROCESSING
            
     Ok-Keun Shin*, Hyun-Ki Kim and Young-Do Chae 
     
Electronics and Telecommunications Research Institute
Korea
(* e-mail okshin@kiet.etri.re.kr)

In this paper, we introduce an experimental hardware 
architecture of a multimedia data processing system called 
ComBiStation, whose applications area includes multimedia 
authoring, CSCW and video conferencing. The platform 
comprises most commonly needed multimedia processing 
functions: audio-visual data capture, playback, multistandard
compression as well as interleaving of compressed audio 
visual data. The proposed architecture minimizes the CPU 
overhead that might be caused by multimedia data processing 
and assures the fluent data flow among system components. We
begin with overall architecture of the whole system, and then 
audio-visual data capture/display unit and multistandard 
compression unit are discussed. Implementation issues and 
future works are also discussed.

Keywords : Multimedia Platform, Compression, CSCW, Video 
Conferencing


DITHERING AS A METHOD FOR IMAGE DATA COMPRESSION

            Pavel Slavik, Jan Prikryl 
            
e-mail slavik@cslab.felk.cvut.cz, prikryl@sgi.felk.cvut.cz

Abstract is not available.

Keywords : data compression, dithering, parallel processing


ASPECT GRAPHS OF THREE_DIMENSIONAL SCENES

            Eduard Sojka 
            
Department of Computer Science
Technical University of Ostrava            
ul.17.listopadu, 708 33 Ostrava-Poruba
Czech Republic
e-mail eduard.sojka@vsb.cz

The paper presentes a new approach to the theory of aspect 
graphs. This approach is based on the presumption that an 
equivalence relation can be introduced on the set of images 
of a given scene. The equivalence relation induces the 
decomposition of the set of images, which is infinite, into
the finite number of classes. Similarly, also the space 
surrounding the scene is decomposed into the classes. From
every point of certain class, equivalent images of the scene 
are perceived. The aspect graph is the graph that contains
the information about these classes and the relations between
them. Presented theory enables to put the approaches 
published earlier into the unifying framework. The paper also
presents an example in which the general theory of aspect
graphs is applied to the three-dimensional problem. Also an
algorithm for computing the decomposition of space into the
classes is proposed.

Keywords : Aspect graph, visual potential graph, visibility


DISCRETE RAY TRACING HIGH RESOLUTION 3D GRIDS

           Nilo Stolte, Rene Caubet 

Institut de Recherche en Informatique de Toulouse
118, Route de Narbonne
31062 - Toulouse - France
e-mail stolte@irit.fr

This article suggests a new approach to discrete Ray-Tracing 
in which a two step three-dimensional DDA and an octree are 
used. A very important problem regarding this kind of 
Ray Tracing, also known as Raster Ray Tracing, is the amount 
of memory required to store the 3D raster grid which will 
contain the discretized scene to be visualized. Since the 
resolutin of this grid is huge because the voxel is assumed 
to be approximatelly the size of a pixel on the screen, it is l
imited by the maximum amount of memory of today's machines. 
Although  using an octree and assuming that the majority of 
space will be empty, as such is the case in most scenes, 
an important memory saving is achieved. This memory saving 
helps using discrete Ray Tracing in normal workstations. It 
is shown that the special octree presented here doesn't slow 
down Ray Tracing significantly and even competes with normal 
discrete Ray-Tracing. It is also shown that in critical cases 
this octree will not occupy much more space than a normal 
3D grid.
	Another important problem is the bottleneck caused by 
the three dimensional DDA in such huge resolution grids. This
problem is solved by dividing these grids into two low 
resolution ones. The process can be divided in two steps 
where optimal times of three-dimensional DDA are achieved. 
This is shown through the comparative tests with the single 
step process.

Keywords : Ray-Tracing, Discrete Ray_tracing, 
Raster Ray-Tracing, Octree, Three-dimensional DDA, Voxel


ON CONVERGENCE AND COMPLEXITY OF RADIOSITY ALGORITHMS

       Laszlo  Szirmay-Kalos and Gabor Marton 
       
Department of Process Control
Technical University of Budapest
Budapest, Müevetem rkp. 11,H-111
Hungary       
e-mail szirmay@fsz.bme.hu

This paper evaluates and compares the convergence and 
complexity characteristics of radiosity algorithms with 
a special emphasis on randomized methods. 

Keywords: Radiosity method, complexity evaluation, 
randomized algorithms


VEGA : VIENNA ENVIRONMENT FOR GRAPHICS APPLICATIONS

      Robert F. Tobler, Helwig Löffelmann, 
                  Werner Purgathofer 

Institute of Computer Graphics
Technical University of Vienna
A-1040 Karlsplatz 13/186/2
e-mail rft@cg.tuwien.ac.at, helvig@cg.tuwien.ac.at,
       wp@cg.tuwien.ac.at

This paper presents a software development environment for 
rendering applications. The main parts of this environment 
are a set of rules concerning the coding style, a set of 
tools to maintain source files and a number of libraries 
providing graphics functionality. The environment has been 
succesfully used in a number of internal projects, dealing 
with rendering.


A ROBUST OBJECT MODELING SYSTEM USING A RANGE FINDER

     Takashi Wada, Tatsuro Nakamura, Kaoru Suzuki 
     
Kansai Research Lab. Toshiba
8-6-26 Motoyoma-Minami-machi, Higashi-Nada-ku, Kobe, 658 Japan

Multimedia Engineering Lab. Toshiba
209, Suhiro-chou, Ome, Tokyo, 198 Japan     

e-mail wada@krl.toshiba.co.jp, 
       tatsuro@mmfive.pms.mmlab.toshiba.co.jp
       suzuki@krl.toshiba.co.jp


This paper presents a robust object modeling system that 
generates an object model from multiviewing range data. The 
purpose of the system is to obtain an exact model, overcoming 
a noise problem in range data. Our system computes object 
existence in voxel space by integrating observations form 
multi-viewpoints. The noise in range data is excluded by this
integration process. Then, a triangular mesh is efficiently 
obtained by using adjacent relationships between voxels. 
Finally, a fine object model is acquired by associating 
brightness values in color image with the vertex on the mesh.
Experiments are performed for plural real objects (dolls and 
a cup) and effectiveness of the system is shown.


FREE FORM SURFACE CONSTRUCTION USING GREGORY PATCHES

             Shouqing Zhang, A.L.Thomas 
             
e-mail: S.Zhang@central.sussex.ac.uk


This paper deals with the construction of a free form surface 
by using both triangular and rectangular Gregory patches. 
The input is a set of 3D points associated with normal values 
at each point. All the control points needed for Gregory 
patch can then be calculated automatically to achieve the G1
continuity between neighbouring patches within surface.

Keywords: Gregory patch, Triangular and Rectangular shape, 
G1 continuity     


EMPLOYING CONSTRAINTS TO FONTS

Borut Zalik, Fiaz Hussain and Nikola Guid 

University of Maribor
Faculty of Technical Scienes
Department of Computer Science
Smetanova 17, SI-62000 Maribor
Slovenia
e-mail zalik@uni-mb.si, fiaz@dmu.ac.uk

This paper gives a new approach for describing the outlines 
of characters. It uses geometric constraints to model 
characteristics of fonts. Such an approach is known to have 
applications in the area of rasterization, where so-called 
hints are introduced to maintain character legibility at low 
resolutions. In order to promote flexibility and performance, 
the technique outlined uses the principle of localised 
constraints. Its basic features are highlighted via its 
application on the Bezier cubic curve.

Keywords : Bezier, constraint-based design, 
feature-based design, constraint solving, font, hints, 
rasterisation.


TEXTURE CLASSIFICATION BY NEURAL NET IN MEDICAL SONOGRAPHY

           V. Cerny, L. Fekete, R. Zajac 
           
Institute of Physics, Comenius University
Mlynska dolina, 842 15 Bratislava, Slovakia

Department of Urology, Medical Faculty Hospital
Hlboka 7, 811 05 Bratislava, Slovakia           
e-mail cerny@fmph.uniba.sk

We have developed a simple neural net classifier to evaluate 
textural informational content in the sonographic medical 
images. The net is trained on a set of texture patterns from 
sonographic images of testes. The samples in the set were 
classified by the supervisor into two classes: "normal" and 
"tumor". After training the performance was 85% correctly 
classified images. We are in the stage of collecting data for
the independent test set of samples.


A NOTE TO THE STRUCTURE OF BRESENHAM SLICE ALGORITHM

               Tomas Hruz, Ivo Povazan 
               

Slovak Technical University
Faculty of Mechanical Engineering
Namestie Slobody 17
812 31 Bratislava, Slovakia               

Institute of Control Theory and Robotics
Slovak Academy of Sciences
Dubravska cesta 9
842 37 Bratislava, Slovakia

e-mail hruz@cvt.stuba.sk, utrrpova@savba.savba.sk

      
This paper gives a complete analysis of the structure of 
segments (spans) of points with a constant y-coordinate, 
which are observed in the Bresenham Slice algorithm. In 
particular, we show that this structure is related to the 
greatest common divisor of the end-point coordinates.


SCIENTIFIC VISUALIZATION WITH MAPLE

           Jiri Hrebicek 
           
Waterloo MAPLE software GmbH, Heidelberg

Abstract is not available.


CATIA SOLUTIONS VERSION 4 - THE OVERVIEW OF THE CAD/CAM/CAE 
                          SYSTEM

                     Ing. Miroslav Kral 

IBM 
Czech Republic

CATIA Solutions Version 4 builds on more than a decade of 
proven marketplace success for CATIA, which currently has 
more than 4 000 customers of all sizes worldwide in 
industries ranging from general manufacturing to aerospace, 
automotive and customer goods. CATIA is internationally 
recognized as an industry leader because it has benefited so 
many users and because of its comprehensive, integrated 
approach to the entire product design, manufacture and 
support process.


REDUCING THE NUMBER OF SHADOW RAYS IN BIDIRECTIONAL PATH 
                        TRACING

            Eric P. Lafortune, Yves D. Willems 
            
Department of Computer Science
Katholieke Universität Leuven  
Celestijnlaan 200A, 30001 Heverlee, Belgium          
e-mail Eric.Lafortune@cs.kuleuven.ac.be

Bidirectional path tracing is a fairly new Monte Carlo 
algorithm for physically based rendering. Introduced as 
a generalisation of path tracing, the algorithm traces paths
for each pixel, not only from the eye point but also from the
light sources. It subsequently links the intersection points
on the respective paths by means of shadow rays. Each shadow 
ray determines if a lighting contribution has to be added to 
the estimate of the radiance of the pixel. 
	In this paper we present a  technique which reduces 
the number of shadow rays that are traced effectivelly. The 
resulting slight increase in the variance of the stochastic 
process can be undone by taking more samples. Practical tests 
show that the approach yields a reduction of the variance for
the same amount of work.


SIMULATING FEATURE-BASED MODELLING IN PROGRAMMING-BY-EAMPLE 
                      PARAMETRIC CAD

       Evgueni N. Loukipoudis and Jan A.A. Melkebeek 
       
Department of Electrical Power Engineering
University of Gent
Sint-Pietersnieuwstraat 41, B-9000 Gent, Belgium       

As an alternative to the development of a full feature-based 
CAD system, we suggest a fairly simple extension to a system 
that possesses parametric capabilities so that feature-based 
modelling interface is simulated. We use 
programming-by-example in defining features by parametric 
programs following a purely procedural approach. The 
parametric programs constitute the feature model which is 
built up in a bottom-up manner. The structuring of the object
in features allows modifications to the parametric 
description in a graphical manner. Some problems related to 
the use of feature in an engineering system for the purpose 
of simulation and analysis as well as the handling of duality 
and comlementarity in a feature model are discussed. We have 
restricted our approach to 2D models and have implemented 
Electric Machinery specific features into 2D CAD system.


    STEREOGRAPHIC VISUALIZATION OF TURBULENT PIPE FLOWS 
      USING ANALYPHS WITH TWOFOLD CENTRAL PROJECTION 

              Wolfgang Messner, Walter Huber 

Technische Universität München
Institut für Informatik
Lehrstuhl für Ingenieuranwendungen in der Informatik
D-80290 München, Germany
e-mail messner,huberw@informatik.tu-muenchen.de

The size of velocity-vector fields obtained from numerical 
solutions of the Navier-Stokes equations for 
three-dimensional, time varying and turbulent flows is 
extremely large. Stereographic systems can intuitively depict
the small depth movements and relations between fluid 
particles that standard perspective projections fail to. As 
expensive high-end-stereographics systems aim at one user one 
computer and are therefore not suitable for demonstration 
purposes in lectures, the anaglyphic method is applied for 
visualizing turbulence in a stereographic movie to a large 
number of people. A solution to overcome the effect of 
ghosting, which has been a major drawback in the use of 
anaglyphs, is introduced. A twofold central projection was 
found to be superior to standard on-axis projections in 
graphics harware allowing an easy representation of objects 
in front of and behind the projection plane and thus 
enhancing the stereoscopis effect. The visualization system 
deals with the small fluctuations of the velocity about its 
mean values (i.e. the so called rms values) to depict real 
turbulence in a pipe flow at reynolds number Re approx.7000.


         SOUND MAPPINGS FOR SURFACE VISUALIZATION

             Rosane Minghim  and A.R.Forrest 
     
University of East Anglia
School of Information Systems
Norwich NR4 7TJ
UK     
e-mail rm@sys.uea.ac.uk, forrest@sys.uea.ac.uk

This paper is concerned with the development of a system to 
support Scientific Visualization by utilising the sound 
channel. The mappings presented here are intended to support 
understanding and interaction with surface data. This is done 
by using sound signals associated with methods of analysis of
surfaces to help the interpretation of surface shape, data, 
and classification. Several examples of sound mappings are 
presented and the implications for data interaction are 
discussed.


     A PARALLEL RADIOSITY SYSTEM FOR LARGE DATA SETS

               David Stuttart, Adam Worrall, 
             Derek Paddon and Claire Willis 

Department of Computer Science
University of Bristol
Bristol, BS8 1TR, UK
e-mail derek@compsci.bristol.ac.uk

The radiosity method gives realistic results for syhtnesising 
images based on diffuse light interaction; however, the 
method is also very computationally expensive. Here we 
present a system for parallelising a progressive refinement 
radiosity method using a ray casting approach. The system 
decomposes the environment into two distinct entities, the 
model geometry and the radiosity sample points. These are 
then distributed amongst the processors and message passing 
communication protocols and caching methods are introduced 
to enable the system to cope with very large models while 
maintaining a high processing rate.


      COMPUTER ANIMATION AND HUMAN ANIMATORS

                 Philip J. Willis 
                 
Computing Group, School of Mathematical Sciences
University of Bath
Claverton Down. Bath BA2 7AY, UK                 
e-mail P.J.Willis@bath.ac.uk

There is often a gap between what technology can deliver and 
what users want. Sometimes this is because of fundamental 
technological constraints but often it is because 
technologists do not address the realities of user needs and
expectations before attempting to solve the problem.
	The talk will describe our approach to computer 
animation and some of the pitfalls of trying to address 
a community which is visually more literate than typical 
computer graphic people, while being far less computer 
oriented.

Keywords : computer animation, animation



LOAD BALANCING FOR PARALLEL RAYTRACER ON VIRTUAL WALLS

       Jiri Zara, Ales Holecek, Jan Prikryl, 
             Jan Burianek, Knut Menzel 
             
e-mail zara@cs.felk.cvut.cz, holecek@sgi.felk.cvut.cz, 
       prikryl@sgi.felk.cvut.cz, xburiane@sgi.felk.cvut.cz,
       knut@uni-paderborn.de

Dynamic load balancing of parallel ray-tracing algorithm 
based on spacial subdivision is discussed. We attempt to find
the load balancing method with the fastest respond on the 
load extreams in the system. An optimal architecture of the
computational system based on three level process hierarchy 
is proposed. The scene partitioning is based on Virtual Walls.

Keywords : computer graphics, rendering, parallel algorithm,
           ray-tracing, load-balancing


GEOMETRIC MODELLING IN A THREE-DIMENSIONAL ENVIRONMENT

            G.J. Clapworthy,  R.A. Noble 
            
Department of Computer & Information Sciences
De Montfort University
Hammerwood Gate
Kents Hill
Milton Keynes, MK7 6HP
e-mail gclapwor@dmu.ac.uk

Computer animators have long had difficulties in defining and
controlling the shape of deformable objects during motion. In 
the context of human figure animation, a new form of surface
description has been developed. Murugaiyan \& Clapworthy [1],
which allows the surface to be manipulated by adjustment of 
a single parameter. The form of the surface deformations 
produced by the operations is under the control of the user.
The mathematical structures used are extensions of NURBS, 
but have a greater flexibility at the cost of 
marginally-increased overheads.
	The development of this tool enables the surface 
modelling of a deformable object to be speeded up 
considerably, and the paper describes a geometric modeller
currently under construction which will use this facility to
produce the desired object by 'virtual sculpting', a process
analogous to the production of real objects from clay. In
this, the user is able to pull and push the surface shape 
to produce either rounded or shaped surface indentations or
protrusions by direct interaction with the surface.
	The object is viewed stereoscopically  and at present
shutter glasses are used. These have the attendant problem of 
ghosting, but this can be removed by a subtract-from-background
algorithm, Noble [2]. Ongoing work at De Monfort university 
[3,4] will eventually allow enhancement using autostereoscopic
displays and eye tracking.
	The object is oriented in three dimensions using 
a spaceball and the surface operations are controlled using 
a mouse. The reduction in the number of paremeters employed 
in the surface description enables the operations on the 
surface to be performed by the use of the mouse buttons. 
Thus, the shape of the object is defined by simple 'point and
click' operations providing an extremely rapid, interactive
method of surface construction.
	It is anticipated that this modelling system could be 
used in a number of areas of work, including Virtual Reality.
	While the tool is still at the developmental stage, 
the paper lays out the complete conceptual basis for 
its construction and implementation.


USING OF DISCRIMINATORY ANALYSIS TO PATTERN RECOGNITION
                OF TACTILE PATTERN
                
                    Jaromir Volf 

Czech Technical University
Faculty of Mechanical Engineering
Technicka 4, 166 07 Prague 6
Czech Republic
Tel.:+422 24 35 27 37, fax +422 24 31 02 92
e-mail volf@fsid.cvut.cz

This papre described a new method of tactile pattern 
recognition. The method can be used a numeric regulator, too. 
This method is based on the discriminatory analysis. The 
computation is going through all system of pattern classes. 
This corresponds to classical computing process, which is 
used in the discriminatory analysis. The Ivanovic deviation 
is used as discriminatory function. The m dimensional vector
is impute and the method can classify into n groups of 
patterns. Pattern recognition can work at a real time.

Keywords : discriminatory analysis, tactile, pattern, image,
           pattern recognition
           

COMPARISON OF SOME RAY GENERATORS FOR RAY TRACING 
                 VOLUMETRIC DATA

                   Milos Sramek 
                   
Institute of Measurement Science
Slovak Academy of Sciences
Dubravska cesta 9
842 19 Bratislava, Slovak Republic
                   
Ray tracing is a popular technique for rendering not only 
analytical objects, but also volumetric data, either 
synthetized or measured. Different techniques for the ray 
definition by traversal of the voxel space were designed. The
first part of the paper is an overview of ray tracing or ray
casting algorithms for the volume data. Stress is laid upon 
exploiting various kinds of speed up techniques. In the 
second part, we compare some of these approaches in 
a computer experiment. Further we show that an appropriate 
choice of background traversal algorithm can significantly 
shorten computation time of the exact ray-surfrace 
intersection point.


          GRAPHICAL  REPRESENTATION OF RULES 
          IN AN OBJECT-ORIENTED ENVIRONMENT

    Kenneth Messa,  Bogdan Czejdo,  Erick Villalobos 
    
Department of Mathematics and Computer Science
Loyola University
New Orleans, LA
U.S.A.

In this paper a graphical representation of rules is 
proposed. The graphical rules are formulated based on 
Extended Entity-Relationship (EER) diagrams. Such diagrams 
are compiled into an object-oriented language code that is 
included as a component of the developed software system. 
Other components can be directly coded in an object-oriented 
language or generated using other graphical tools. We 
identify and discuss two types of graphical rules : class 
graphical rules and instance graphical rules. The syntax and 
semantics of both types of graphical rules are discussed. The
graphical rules can contain object-oriented code in addition 
to graphical diagrams. The methods of integration of diagram
and object-oriented code are described.