Members of the Computer Graphics Group
have participated in the following projects :
Three Dimensional Object Retrieval
(Ministry of Education research grant, THALES)
2011-2015
Lead Principal Investigator (Democritus University of Thrace): I. Pratikakis
Principal Investigator
(UoA) : T.Theoharis
PhD Researchers : A. Danelakis, K. Sfikas, P. Theologou
Budget : € 500.000
The research focused on three aspects of 3D Object Retrieval:
(i)
Retrieval of 3D Objects by using Global Shape Descriptors: methods for
the alignment of 3D objects (emphasizing on the retrieval), rigid 3D
object descriptors and finally efficient methodology for the retrieval
of 3D objects with 2D query. (ii) Retrieval with Segmented Shape
Descriptors: the possibility of using, in addition to the shape
information, also the structure of an object, as this is returned after
a segmentation process. (iii) Retrieval from 3D Video Databases: 3D
faces and dynamic 3D human behavior, in both cases the extraction of
spatio-temporal descriptors and the efficient matching over time was in
focus.
Three Dimensional Within-Class
Object Retrieval with Applications in Ear Recognition
(GSRT research grant - Joint with CBL, University of Houston, TX)
2006-2008
Principal Investigator
(UoA) : T.Theoharis
Principal Investigator (UH) : I. Kakadiaris
PhD Researchers (UoA) :
TBD
Budget : € 60.000 (UoA) + €
60.000 (UH)
As the size of the available collections of 3D objects grows, data-base like operations become essential for their management, with the key operation being retrieval (query). Large collections are also pre-categorized into classes, so that a single class contains objects of the same type (e.g. human faces). Within-class retrieval methods differ significantly to those for intra class retrieval, since they must provide higher discrimination power. At the same time they can exploit the basic characteristics of the object class. The aim of this research is within-class 3D object retrieval, able to discriminate objects with subtle differences. As testing ground, it is intended to use the biometric field of ear recognition. The three dimensional form of the human ear provides a unique identifying characteristic for humans (like the fingerprint or the face). However it is non-intrusive (in contrast to fingerprint capture) and more reliable than facial scans (the ear is more rigid than the facial region: there are no 'ear expressions'). Ear recognition is a novel biometric at an infant research stage and no significant ear databases exist; for this reason we propose to create an ear database and make it available to the research community.
Advanced Methods of Visual Computing
in Cultural Heritage
(GSRT PENED research grant #70/3/8405 - Joint with IPTHL Thessaloniki)
2005-2008
Principal Investigator
(IPTHL) : M. Strintzis
Principal Investigator
(UoA) : T.Theoharis
Senior Researchers (UoA) :
G.Papaioannou, V. Drakopoulos, I. Kakadiaris, N. Paragios
PhD Researchers (UoA) : G.Passalis, P.Stavrou, P. Manousopoulos
Budget : €
224.970
This project aims to apply visual computing techniques in the field of cultural heritage. In particular, it aims to reconstruct and visualize cultural entities. This process is divided into four stages: a) 3D object retrieval in digital libraries based on material similarity, age, extracted resemblance (patterns, carvings) or even complementarity of characteristics such as texture or morphology. b) Synthesis (part matching and combination). c) Filling (repair of incomplete surfaces and holes). d) Real time realistic and animated visaulization of the reconstructed object or scene.
Knowledge Rased Retrieval of 3D
Models
(GSRT PENED research grant #70/3/8419 - Joint with EKEFE Demokritos)
2005-2008
Principal Investigator
(Demokritos) : S. Perantonis - I. Pratikakis
Principal Investigator
(UoA) : T.Theoharis - S. Theodoridis
Senior Researchers (UoA) :
V. Drakopoulos
PhD Researchers (UoA) : P. Papadakis
Budget : €
107.400
The main goal is the
investigation of novel representation techniques for 3D models with the
aim of efficient 3D object retrieval in multimedia databases.
Computer Aided Restoration of
Archaeological
Monuments using Computer Graphics
(University of Athens
research
grand #70/4/3241)
1997-1998
Principal Investigators :
A.Boehm
Members :
Th.Theoharis,
G.Papaioannou, E.A.Karabassi
In this project, an
effort
was conducted to model and solve the problem of the reconstruction of
the
archaeological site of Athens' Acropolis with the aid of computer based
methods. The aim was not the creation of another tool for a virtual
reconstruction
of the site based on architectural plans, but the efficient
representation
of archaeological data (digitized archaeological fragments) and
the
development of algorithms for automatic measurement, classification and
combination of the available data in order to assist the work of the
Acropolis
restoration group.
Part
of the work related to the archaeological fragments representation,
focuses
on the modeling and exploitation of texture information, as described
in
"G. Papaioannou, T. Theoharis, A. Boehm, A Texture Controller,
The
Visual Computer, Springer-Verlag, 14(10), pp 488-496, 1998".
Computer Aided Simulation and
Graphical
Representation of Accumulation Phenomena
(University of Athens
research
grand #70/4/3241)
1998-1999
Principal Investigators :
Th.Theoharis
Members :
E.A.Karabassi,
G.Papaioannou
The goal of this
project
was to study the representation of accumulation phenomena, e.g.
accumulation
of dust, snow, mud etc. The project follows a "computer graphics
approach"
of the subject, where the goal is to achieve realistic images
incorporating
such phenomena, rather than focusing on a simulation from the
physicist's
point of view. Research was basically conducted on fast and accurate
collision
detection algorithms between particles and surfaces.
In most particle systems particle dimensions are very small relative to
the remaining objects; therefore collision detection between particles
and objects is reduced to simple ray/surface intersection tests.
However
this is not always a valid assumption : e.g. in cases of materials like
slime, mud, or snow the volume of the particles may not be ignored, as
this would not only affect physical simulations but could also lead to
non-realistic images. In order to study the behavior of such
models,
where particles are manipulated as spherical objects of a non
negligible
volume, we have developed an intersection detection algorithm between
spherical
particles’ trajectories ( i.e. cylinders with spherical caps ) and
polygons.
Our algorithm consists of a set of intersection tests between
spheres/cylinders
and lines/triangles.
The method is discussed
in "E.A. Karabassi, G. Papaioannou, Th. Theoharis, Intersection Test
for Collision Detection in Particle Systems, ACM Journal of
Graphics
Tools, (accepted for publication)".
Multiresolutional Editing of Polygon
Models
(University of Athens
research
grand #70/4/4151)
1998-1999
Principal Investigators :
F.Tzaferis
Members : N.Platis
Multiresolution
analysis
techniques provide efficient ways of displaying complex scenes, by
storing
multiple levels of detail of the scene and displaying each time only
the
required amount of detail for visually acceptable results. This project
concentrates on multiresolution methods that deal specifically with
polygonal
models. An interesting extension of these methods is to support
multiresolution
editing of the models: the user can work in coarse resolutions to
change
the overall shape of the model, or in more fine resolutions for the
details;
in any case, the full multiresolution representation of the model must
be updated to reflect the changes.
The
project
evaluates the existing multiresolution methods for their support of
editing.
The most suitable ones will be extended as needed and editing will be
implemented
for the best of them.
Fractals and Dynamic Systems
(University of Athens
research
grand #70/4/1388)
Principal Investigators :
Members :
Mandelbrot-like Sets of Konig
Iteration
Functions
(University of Athens
research
grand #70/4/2403)
Principal Investigators :
Members :
Fractal Interpolation Functions
(University of Athens
research
grand #70/4/3247)
Principal Investigators :
Members :
Fractal Interpolation Surfaces
(University of Athens
research
grand #70/4/4160 )
Principal Investigators :
Members :
PATRA
(EPET II / PAVE)
Computer Graphics Group
members : E.A.Karabassi, G.Papaioannou
The objective of the PATRA
project is the development and clinical evaluation of an industrial
prototype
of a system used in cooperation with the existing angiographic
equipment
(add-on), which will allow the reconstruction, visualization and
processing
of a 3D model of the cardiac structure under examination. In other
words,
3D modeling of the human vascular system. We have developed
specifically
for this project an angiography processing subsystem named
Angioenhancer
to semi automatically extract vessel structures from digital
angiography
data.
PATRA home page
3DHeartView
(ESPRIT 24484)
Computer Graphics Group
members : G.Papaioannou, E.A.Karabassi
3DHeartView
home page