CGiV2010 -
7th International Conference
Computer Graphics, Imaging and Visualization
7, 8 - 10 August 2010
· University
of Technology, Sydney · Sydney · Australia ·
http://www.graphicslink.co.uk/cgiv2010/
|
|
|
1. Non-rigid registration of
partial 3D scans Hamid Laga, Assistant Professor,
Global Edge Institute, Tokyo Institute of Technology, Tokyo, Japan Zulfiqar Habib, Associate Professor,
Department of Computer Science, National University of Computer &
Emerging Sciences (FAST), Lahore, Pakistan  |
A half-day Course:
Saturday 7th August 2010, Time: 10:00 -13:00
Non-rigid registration of
partial 3D scans
Hamid
Laga, Assistant Professor, Global Edge Institute,
Tokyo Institute of Technology, Tokyo, Japan
hamid@img.cs.titech.ac.jp
http://www.img.cs.titech.ac.jp/~hamid
Abstract
Nowadays, a large number
of interesting applications require high quality surface models. A number of
different methods have been developed to perform the acquisition from real
world objects. Existing scanning devices however provide only partial views of
the geometry at each time frame with no correspondence information. Obtaining a
complete representation of the acquired object requires the alignment and
merger of the partial surface data provided by the sensors of the scanning
system. This problem, known as registration, has been solved for the case where
the object stays rigid. However, applications such as marker-less motion
capture, modeling animated characters, creating special effects for films, or
3D video, deal with objects that move and deform during the acquisition
process. Scanning such dynamic and time-varying surfaces, i.e., objects that
cannot remain static for an extended period of time, is still an open
challenging problem.
This tutorial aims at
reviewing the basic techniques and the recent advances in partial scan
registration with a special focus on deformable and time-varying surfaces.
Brief description of
tutorial’s organization:
The first part of the
tutorial gives an introduction to non-rigid registration of deformable and
time-varying surfaces for attendees with a basic geometry processing and
computer vision background. In the second part we focus on the current
state-of-the art techniques. The third part will focus on the use of non-rigid
registration in computer vision and computer graphics applications such as
performance capture and 3D video. The last part discusses the open research
problems and aims at stimulating young researchers to join the research
community.
The structure and time
allocation of the main sections of the tutorial are as follows:
1. Introduction (10 min)
2. Rigid registration (15 min)
a. Basic ICP Algorithm.
b. ICP variants.
3. Non-rigid registration (60 min)
a. Formulation
i.
Non-rigid registration as an optimization problem
ii. Probabilistic formulation
b. Procedure
i.
Feature detection and selection
ii. Correspondence computation
c. Issues
i.
Motion constraints (isometric deformations vs. free
deformations).
ii. Large scale motion vs. detail synthesis
d. Summary
4. Non-rigid extensions of the ICP algorithm
(15min)
5. The state-of-the art techniques (30 min)
a. We review the recently published work on
non-rigid registration.
6. Applications (25min)
7. Summary and discussion (25min)
Organisation
The
total duration:
Half day (3 hrs).
Level
of Tutorial:
beginners with basic geometry processing and computer vision background.
Hamid Laga is
an Assistant Professor at the Global Edge Institute of Tokyo Institute of
Technology since Oct. 2006. From April 2006 to September 2006 he was a JSPS
postdoctoral fellow at the Image Processing Laboratory of Nara Institute of
Science and Technology (NAIST). He received his PhD and M.S degrees in March
2006 and March 2003, respectively, from the Computer Science Dept. of Tokyo
Institute of Technology. His research interests include object detection and
tracking, pattern recognition, shape analysis, multimedia information
retrieval, digital geometry processing, and high performance computing for
vision and graphics. He has published several journal and conference papers on
3D object retrieval and people tracking. He is the recipient of the Best Paper
Award at IEEE International Conference on Shape Modeling and Applications 2006,
the Best Paper Award at NICOGRAPH 2007 paper contest, and the Best Paper Award
at the International Computer Graphics Paper Grand Prix, the Society for Art
and Science of Japan. museum curation, the
relationship between text and image in medieval art, and the artistic origins
of information visualization.
A half-day Course:
Saturday 7th August 2010, Time: 14:00 -18:00
Zulfiqar
Habib, Ph.D., Associate Professor, Department of
Computer Science
National University
of Computer & Emerging Sciences (FAST), B-Block, Faisal Town, Lahore 54590.
E-mail:
zulfiqar.habib@nu.edu.pk; drzhabib@hotmail.com
Abstract
Path planning is one of
the fundamental problems, with numerous applications in the fields of science,
engineering and technology such as networks, GIS, CAD systems, environmental
design, digital image processing, communications, geometric modeling, and other
disciplines. Particularly fair curves are used in highway or railway designing,
robotics, collision detection and avoidance, graphical animation, power supply,
water supply, fiber optics deployment, digital data compression, and gear
designing. Path planning is a mechanism for deciding which path is the best to
follow on the basis of terms expressed in current internal representations of
vertical dimension of the surface. In path planning it is often required to
pass a fair curve through given coordinates or consecutive points. A fair curve
should not have unwanted loops, cusps, inflection points, and curvature extrema. Fair path with obstacle avoidance also needs to be
shortest for the purposes of time, material, or energy saving. One of the main
approaches to path planning is through the use of spline
functions. This tutorial aims to deliver the basic, practical, and state of the
art techniques of fair path planning. It concludes with the introduction of
industrial projects and open research issues to motivate young researchers to
join the research community. This tutorial is designed for beginners and
suitable for both undergraduate and graduate students looking for new research
ideas and directions.
Brief
description of tutorial’s organization
This tutorial is a common topic of the following subjects:
- Data
Visualization
- Geometric
Modeling
- Computer
Graphics
- Numerical
Methods
It is organized in the following sections:
- Introduction
- History
- Applications
- Path planning requirements
- Physical problem
- Shortest & smooth path of a robot arm
- Bezier curves
- Pyramid algorithms
- Geometric construction
- Properties
- Bernstein polynomials
- Spline
- Parametric continuity
- Geometric continuity
- Parametric cubic spline
- Solution of physical problem
- Industrial projects
- Fair path planning with spiral transitions
- Latest research issues
- Open research issues
- Questions/Answers and discussion
The
total duration:
Half day (4 hrs).
Level
of Tutorial: Beginners
(No Pre-req.)
Dr. Zulfiqar Habib earned his Ph.D. degree in Computer
Science in 2004 from Kagoshima University Japan followed by the award of post
doctoral fellowship of two years by Japan Society for the Promotion of Science.
Currently he is holding the position of Associate Professor at National
University of Computer & Emerging Sciences (FAST), Lahore. Dr. Habib has
achieved various awards in education and research including three graduate
merit fellowships by Japanese and German Governments. He is an associate editor
of the International Journal of Computer Graphics & CAD/CAM, reviewer of
many famous journals and is organizer of various international conferences. He
has published a large number of high impact factors and well cited papers.
Registration-Form | TOP | Course – Tutorial Submission