This tutorial provides an introduction to the latest active 3D sensing techniques found in the literature and in commercial products. Although the emphasis is focused on active triangulation and time of flight systems, other techniques such as optical interferometry and passive methods, are discussed. For each type of 3D method, optical, hardware, and algorithmic related topics as well as case studies are presented. Basic knowledge of 3D measurement principles, Gaussian beam propagation and optical signal detection with solid-state detectors is provided so that critical aspects like image resolution and range image precision can be evaluated. Furthermore, algorithms related to polychromatic reflectance estimation, control of scanning mechanisms, calibration, and some key applications are presented. Among the applications presented, some were demonstrated by NRC researchers and others by our collaborators. In particular, it is shown that active 3D sensing is a powerful documentation tool that can complement other traditional methods. For instance, the SIBA coordination of the University of Lecce uses digital 3D imaging in the context of heritage documentation as a complementary service for their current digital library services.

Benefits
The aim of this tutorial is to provide the attendees with tools to evaluate and better understand active 3D sensing techniques for their own applications. Those applications may include building or purchasing a 3D system and understanding the nature of the 3D images in order to apply optimal geometric processing algorithms. This tutorial should help the attendee in modeling a given sensor into a "black box", and in understanding image artifacts that can be encountered. The case studies will expose the attendee to some of the most interesting applications of active 3D sensing. These applications will in turn demonstrate that the technology is available and can be used by everyone.

Intended audience
People who are involved in the development, evaluation and application of 3D sensing techniques in an industrial, commercial or educational context.

J.-Angelo Beraldin joined the National Research Council of Canada in 1986.
His current research interests include sensor systems engineering and signal
processing (hardware and software) for 3D vision systems. He has developed
expertise in opto-electronic aspects of range sensors as well as in the
intrinsic/extrinsic calibration of them. He has more than 70 papers and
publications, and, has licensed technology to various Canadian companies.
Mr. Beraldin has presented the tutorial on active 3D vision in many
occasions throughout North America and Europe. He is very active in the
field of heritage in Italy for which he organized three workshops.  He is
also a registered professional engineer with the province of Ontario.

Francois Blais is a Senior Research Officer at the National Research Council
of Canada. Since 1984, he has been involved in the development of many of
the different 3D technologies at NRC, especially their range sensing
systems, and their applications. He has also been acting as Director of R&D
for Vitana Corp. to supervise the companies R&D efforts in 3D. Since April
1999, he returned at the NRC to continue his research.  His topics of
interest cover various fields in digital and image processing, control, 3D
vision systems and their applications. He has more than 75 papers and
publications, and 8 patents of which several have been licensed to different
Canadian industries. Mr. Blais is a registered professional engineer of
Ontario.

Virginia Valzano is the general director of the SIBA Coordination (Servizi
Informatici Bibliotecari di Ateneo) and of the Telematic Information System
for Research and Education of Lecce University. She is also professor at the
Faculty for Heritage at the University of Lecce in "Organizzazione informatica delle biblioteche" and in "Teoria e tecniche della catalogazione e classificazione".
She is in charge of many national projects in Italy and she acts as a
scientific advisor in a number of national and international projects aimed
at the preservation, enhanced knowledge, and the assessment of valuable
objects of high historical and artistic value including archeological sites.
Virginia Valzano has many publications in her field of expertise, i.e.,
information science and digital technologies as applied to heritage.
 

A description of how one can represent high level knowledge in a manner compatible with lower level will be addressed. He will describe how some procedural systems calling low level or intermediate level routines can interrogate and reason about the low level knowledge. Emphasis will be placed on the importance of maximum likelihood reasoning to produce a system capable of dealing with real-world situations such as false knowledge and missing information. He will also focus on feedback expectancy methods to stabilize the analysis process. During this tutorial, experimental results and an analysis of the pros and cons of each method will be presented.

Benefits

This course will enable the attendee to grasp the basic concepts allowing the extraction of geometric primitives from 3D data. Intended audience

Intended audience

This tutorial is primarily intended for practitioners of computer vision who are interested in state-of-the-art range image processing and understanding. The tutorial requires some basic knowledge of computer vision, differential geometry, and statistics, but these are not prerequisite.

Dr. Pierre Boulanger is a Senior Research Officer at the National Research Council, specializing in the field of range image processing, geometrical modeling, and virtualized reality. He has published more than 50 papers, in journals and for conferences, on various topics relating to range image processing and its applications. Dr. Boulanger has a Ph.D. from the University of Montreal in Electrical Engineering.

The study of human body measurement (anthropometry) using one-dimensional (1D) technology has been around for centuries and researchers are very comfortable with using this technology. As a result three-dimensional (3D) scanning technology is often viewed as a means to obtain these 1D measurements rather than as a new type of measurement capability.

This tutorial provides an introduction to effective engineering anthropometry practices with emphasis on the advantages and needs for using 3D scanning and analysis tools in modeling and design. Common practices in statistics and modeling will be reviewed and effective practices will be presented with a comparison of using traditional 1D measurements to 3D. New capabilities available with the new technology will also be presented.
 

Kathleen Robinette has been working in this field since 1978 and has been with the U.S Air Force Research Laboratory since 1982. She is the Senior Research Anthropologist and Director of the Computerized Anthropometric Research and Design (CARD) Laboratory. She has been focused on 3D anthropometric research since 1986, and is the director of the CAESAR project, a 3D anthropometry survey of North America and Europe, the first of its kind.
 
 
 
 

Séance # 4: 3D Modeling from Images

Marc Pollefeys (Katholieke Universiteit Leuven, Belgium)

This tutorial describes how 3D models can be obtained from images acquired with a hand-held camera. The course will explain the different components which are required to build an automatic system to do so. The course will cover automatic feature matching, projective reconstruction, self-calibration, dense matching and 3D modeling. It will be illustrated with examples ranging from archaeology and architecture to planetary exploration.

Benefits

This course will enable attendees to understand how 3D models can automatically be obtained from images.

Intended Audience

This course is intended for computer vision and computer graphics researchers who are interested in state-of-the-art methods for obtaining 3D models from images. A basic knowledge of computer vision and 3D geometry is usefull but not required.
 
 

Dr. Marc Pollefeys is a Postdoctoral Fellow of the Fund for Scientific Research - Flanders associated with the K.U.Leuven. His research interest includes 3D modelling from images,  self-calibration, multi-view geometry, plenoptic modelling, virtual and augmented reality. He has more than 40 papers and
publications and won several prizes for his research in 3D modelling from images. He is now involved in several projects, ranging from virtual archaeology to stereobased mars rover control. Dr. Pollefeys has a Ph.D. in Electrical Engineering from the University of Leuven.

Takeo Kanade, Carnegie Mellon University, USA (Conférencier invité)

Séance # 1 : 3D Sensors

09h10 – 09h35 A Self-Referenced Hand-Held Range Sensor

Patrick Hébert, Computer Vision and Systems Laboratory, Department of Electrical Engineering, Laval University, Canada
 
 

09h35 – 10h00 Calibration-Free Approach to 3D Reconstruction Using Light Stripe Projections on a Cube Frame

Chang Woo Chu and Soon Ki Jung, Virtual Reality Laboratory, Department of Computer Engineering, Kyungpook National University, Daegu, Korea

10h00 – 10h25 Approaches to a Color Scannerless Range Imaging System, Lawrence A. Ray, Kenneth Repich and Louis Gabello, Eastman Kodak Company, USA

10h25 – 10h55 Pause-café

Séance # 2: 3D Sensors

10h55 – 11h20 Compact and Portable 3D Camera for Space Applications ,
É. Harvey, M. Arsenault, J.-F. Lavoie, B. Bélanger and M.-A. Boucher, Institut national d’optique, Sainte-Foy (QC), Canada

11h20 – 11h45 In Process 3D-Sensing for Laser Material Processing

P. Klinger, B. Spellenberg, J. M. Hermann and G. Haeusler, Chair for Optics, University of Erlangen-Nuernberg, Erlangen, Germany

11h45 – 13h20 Déjeuner

Séance # 3: 3D Sensors

13h20 – 13h45 3D Modeling of Archaeological Vessels Using Shape from Silhouette

Srdan Tosovic and Robert Sablatnig, Vienna University of Technology, Institute of Computer Aided Automation, Pattern Recognition and Image Processing Group, Vienna, Austria

13h45 – 14h10 Stroboscopic Stereo Rangefinder

Jean-Christophe Nebel, Francisco J. Rodriguez-Miguel and
W. Paul Cockshott, University of Glasgow, UK

14h10 – 14h35 Dual-Beam Structured-Light Scanning for 3D Object Modeling

Johnny Park, Guilherme N. DeSouza and Avinash C. Kak, Purdue University, USA

14h35 – 15h05 Pause-café

Séance # 4: 3D Sensors

15h05 – 15h30 Projective Structure and Motion from Two Views of a Piecewise Planar Scene

Adrien Bartoli, Peter Sturm and Radu Horaud, INRIA, France

15h30 – 15h55 A Laser Range Scanner Designed for Minimum Calibration Complexity

James Davis and Xing Chen, Stanford University, USA

15h55 – 16h20 3D Rendering With an Amateur Digital Camera

Magdalena Urbanek, Radu Horaud and Peter Sturm, INRIA Rhone-Alpes, France

16h20 – 16h45 A 3D Laser Micro-Sensor Integrating Control and Data Processing in an FPGA-Based Calculator

Nestor Arana, Maurice Briot, Christian Ganibal, Alexeandre Nketsa, and Roland Prajoux, LAAS-CNRS, France

16h45 – 17h20 Présentation d’affiches (présentation de 2 minutes par affiche)

17h20 – 18h30 Réception et séance d’affichage
 
 

Le mercredi 30 mai 2001

07h30 – 17h30 Inscription

08h30 – 09h10 Modeling from Reality

Katsushi Ikeuchi, University of Tokyo, Japan (Conférencier invité)

Séance # 5: View Planning

09h10 – 09h35 View Planning with a Registration Constraint

William R. Scott, Department of Electrical Engineering, University of Ottawa and Visual Information Technology Group, National Research Council Canada, Ottawa, Canada

Gerhard Roth, Visual Information Technology Group, National Research Council Canada, Ottawa, Canada

Jean-François Rivest, Department of Electrical Engineering, University of Ottawa, Canada

09h35 – 10h00 Computing Camera Positions from a Multi-Camera Head

Gerhard Roth, Visual Information Technology Group, National Research Council Canada, Ottawa, Canada

10h00 – 10h30 Pause-café

Séance # 6: View Registration

10h30 – 10h55 Efficient Variants of the ICP Algorithm

Szymon Rusinkiewicz and Marc Levoy, Stanford University, USA

10h55 – 11h20 Automatic 3D Modeling Using Range Images Obtained from Unknown Viewpoints

Daniel F. Huber, Robotics Institute, Carnegie Mellon University, USA

11h20 – 11h45 A Nearest Neighbor Method for Efficient ICP

Michael Greenspan and Guy Godin, National Research Council Canada, Ottawa, Canada

11h45 – 13h20 Déjeuner

Séance # 7: View Registration

13h20 – 13h45 Robust Recognition and Pose Determination of 3D Objects Using Range Images in Eigenspace Approach

Danijel Skocaj and Ales Leonardis, Faculty of CIS, University of Ljubljana, Slovenia

13h45 – 14h10 A Method for the Registration of Attributed Range Images

Guy Godin, National Research Council Canada, Ottawa, Canada

Denis Laurendeau and Robert Bergevin, Laval University, Canada
 
 

14h10 – 14h35 Reconstruction of Complex Environments by Robust Pre-Aligned ICP

Vittorio Murino, Luca Ronchetti and Umberto Castellani, Dipartimento Scientifico e Tecnologico, University of Verona, Italy

14h35 – 15h05 The Parallel Iterative Closest Point Algorithm

Christian Langis, Michael Greenspan and Guy Godin, National Research Council Canada, Ottawa, Canada

15h05 – 15h30 Pause-café

Séance # 8: Geometric Signal Processing

15h30 – 15h55 Reliable 3D Surface Acquisition, Registration and Validation Using Statistical Error Models

J. Gühring, Institute for Photogrammetry, University of Stuttgart, Germany

15h55 – 16h20 Affine Transformations of 3D Object Represented with Neural Networks

Emmanouil Piperakis and Itsuo Kumazawa, Department of Computer Science, Tokyo Institute of Technology, Tokyo, Japan

16h20 – 16h45 Real-Time Geometrical Tracking and Pose Estimation Using Laser Triangulation and Photogrammetry

F. Blais , J.-A.Beraldin, S. El-Hakim and L. Cournoyer, Institute for Information Technology, National Research Council Canada, Ottawa, Canada

18h30 – 21h00 Banquet - Hôtel Loews Le Concorde
 
 

Le jeudi 31 mai 2001

07h30 – 17h30 Inscription

08h30 – 09h10 How Much 3D-Information can we Acquire?

Gerd Häusler, University of Erlangen-Nürnberg, Germany (Conférencier invité)
 
 
 
 

Séance # 9: Geometric Signal Processing

09h10 – 09h35 Comparison of HK and SC Curvature Description Methods

Helmut Cantzler and Robert B. Fisher, University of Edinburgh, UK

09h35 – 10h00 Fast Range Image Segmentation by an Edge Detection Strategy

Angel Sappa and Michel Devy, LAAS-CNRS, France

10h00 – 10h25 Shape Recovery and Analysis of Large Screw Threads

C. Robertson and R. B. Fisher, Vision Group, IPAB, Division of Informatics, University of Edinburgh, UK

10h25 – 10h55 Pause-café

Séance # 10: Object Modeling

10h55 – 11h20 3D Reconstruction from Two Orthogonal Views Using Simulated Annealing Approach

Jing Ning, Sally McClean Faculty of Informatics, University of Ulster at Coleraine, Co. Londonderry

Kieran Cranley, Radiation Protection and Imaging Section Northern Ireland Regional Medical Physics Agency Forster Green Hospital, Belfast, Ireland

11h20 – 11h45 Processing Range Data for Reverse Engineering and Virtual Reality

S. Karbacher, X. Laboureux, N. Schön and G. Häusler, Chair for Optics, Physics Institute, University of Erlangen, Germany

11h45 – 13h20 Déjeuner

Séance # 11: Object Modeling

13h20 – 13h45 A Method of Style Discrimination of Oil Painting Based on 3D Range Data

Naoya Masuda, Kunihito Kato and Kazuhiko Yamamoto, Gifu University, Japan

Hideki Tanahashi, Softopia Japan Foundation, HOIP, Japan

13h45 – 14h10 Estimation of Elastic Constants from 3D Range-Flow

Jochen Lang and Dinesh K. Pai, Laboratory of Computational Intelligence, Department of Computer Science, University of British Columbia, Vancouver, Canada

14h10 – 14h35 Tolerance Control with High Quality 3D Data

F. Prieto, Universidad Nacional de Colombia Sede Manizales Colombia

T. Redarce, Laboratoire d'Automatique Industrielle INSA de Lyon France

P. Boulanger, Institute for Information Technology, National Research Council Canada, Ottawa, Canada

R. Lepage, LIVIA, École de Technologie Supérieure Montréal, Canada

14h35 – 15h05 Pause-café

Séance # 12: Environment Modeling

15h05 – 15h30 Reconstructing Urban 3D Model Using Vehicle-Borne Laser Range Scanners

Huijing Zhao and Ryosuke Shibasaki, Center for Spatial Information Science, University of Tokyo, Japan

15h30 – 15h55 AVENUE: Automated Site Modeling in Urban Environments

Peter K. Allen, Ioannis Stamos, Atanas Gueorguiev, Ethan Gold and Paul Blaer, Department of Computer Science, Columbia University, USA

15h55 – 16h20 Acquisition of Three-Dimensional Information in Real Environment By Using Stereo Omni-Directional System (SOS)

H. Tanahashi, Y. Niwa, Softopia Japan and JST, Japan

D. Shimada and K. Yamamoto, Gifu University, Japan

Le vendredi 1^er juin 2001

08h30 – 12h00 Inscription

08h30 – 09h10 3D Imaging for Computer Graphics (cg) Applications

Holly Rushmeier, IBM TJ Watson Research Center, USA (Conférencière invitée)

Séance # 13: Human Modeling

09h10 – 09h35 3D Modeling System of Human Face and Full 3D Facial Caricaturing

Takayuki Fujiwara and Hiroyasu Koshimizu, SCCS, Chukyo University, Japan

Kouta Fujimura, Hitoshi Kihara, Yoshiaki Noguchi and Naoya Ishikawa, SANYO Electric Co. Ltd, Japan

09h35 – 10h00 3D Landmark Detection and Identification in the CAESAR Project

Dennis Burnsides and Mark Boehmer, Sytronics, Inc., Beavercreek, USA

Kathleen Robinette, Air Force Research Laboratory, Wright-Patterson AFB, USA

10h00 – 10h25 A Physically-Based Model for Real-Time Facial Expression Animation

Yu Zhang, Eric Sung and Edmond C. Prakash, School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore

10h25 – 10h55 Pause-café

10h55 – 11h20 Plenary Session on Future Trends of 3D Digitizing

11h20 – 11h45 Plenary Session on Applications of 3D Digitizing and Modeling

11h45 – 12h00 Mot de la fin