Template based shape processing [Elektronische Ressource] / Carsten Stoll

Sujets

• Informatik

Informations

• Publié par: universitat_des_saarlandes
• Publié le: 01 janvier 2009
• Nombre de lectures: 23
• Langue: English
• Poids de l'ouvrage: 38 Mo

Extrait

Template based shape processing
Carsten Stoll
Max-Planck Institut Informatik
Dissertation
zur Erlangung des Grades des
Doktors der Ingenieurswissenschaften (Dr.-Ing.)
der Naturwissenschaftlich-Technischen Fakult¨aten
der Universit¨at des Saarlandes
November 2009Date of Colloquium:
30. September 2009
Dean:
Prof. Dr. Joachim Weickert
Faculty of Mathematics and Computer Science
Saarland University
Saarbruc¨ ken, Germany
Examination Board:
Prof. Dr. Philipp Slusallek (Chair)
Saarland University
Prof. Dr. Hans-Peter Seidel
Max-Planck Institut Informatik
Dr. Christian Theobalt
Max-Planck Institut Informatik
Prof Dr. Stefan Gumhold
Technische Universit¨at Dreseden
Dr. Thorsten Thorm¨ahlen
Max-Planck Institut Informatik
iiAbstract
As computers can only represent and process discrete data, informa-
tion gathered from the real world always has to be sampled. While it
is nowadays possible to sample many signals accurately and thus gen-
erate high-quality reconstructions (for example of images and audio
data), accurately and densely sampling 3D geometry is still a chal-
lenge. Thesignalsamplesmaybecorruptedbynoiseandoutliers,and
contain large holes due to occlusions. These issues become even more
pronounced when also considering the temporal domain. Because of
this, developing methods for accurate reconstruction of shapes from
a sparse set of discrete data is an important aspect of the computer
graphics processing pipeline.
In this thesis we propose novel approaches to including semantic
knowledge into reconstruction processes using template based shape
processing. We formulate shape reconstruction as a deformable tem-
plate fitting process, where we try to fit a given template model to
the sampled data. This approach allows us to present novel solutions
to several fundamental problems in the area of shape reconstruction.
We address static problems like constrained texture mapping and se-
mantically meaningful hole-filling in surface reconstruction from 3D
scans, temporal problems such as mesh based performance capture,
and finally dynamic problems like the estimation of physically based
material parameters of animated templates.
iiiKurzfassung
Analoge Signale mussen¨ digitalisiert werden um sie auf modernen
Computern speichern und verarbeiten zu k¨onnen. Fur¨ viele Signale,
wie zum Beispiel Bilder oder Tondaten, existieren heutzutage effek-
tive und effiziente Digitalisierungstechniken. Aus den so gewonnenen
Datenk¨onnendieursprunglic¨ henSignalehinreichendakkuratwieder-
hergestelltwerden. ImGegensatzdazustelltdaspr¨aziseundeffiziente
DigitalisierenundRekonstruierenvon3D-odergar4D-Geometrieim-
mer noch eine Herausforderung dar. So fuhren¨ Verdeckungen und
Fehlerw¨ahrendderDigitalisierungzuL¨ochernundverrauschtenMeß-
daten. Die Erforschung von akkuraten Rekonstruktionsmethoden fur¨
diese groben digitalen Daten ist daher ein entscheidender Schritt in
der Entwicklung moderner Verarbeitungsmethoden in der Computer-
grafik.
In dieser Dissertation wird veranschaulicht, wie deformierbare geo-
metrische Modelle als Vorlage genutzt werden k¨onnen, um seman-
tische Informationen in die robuste Rekonstruktion von 3D- und 4D-
Geometrieeinfließenzulassen. Dadurchwirdesm¨oglich,neueL¨osungs-
ans¨atze fur¨ mehrere grundlegenden Probleme der Computergrafik zu
entwickeln. So k¨onnen mit dieser Technik L¨ocher in digitalisierten 3D
Modellen semantisch sinnvoll aufgefullt,¨ oder detailgetreue virtuelle
Kopien von Darstellern und ihrer dynamischen Kleidung zu erzeugt
werden.
ivAcknowledgements
First, I would like to thank my supervisor Prof. Dr. Hans-Peter
Seidel who made it possible for me to work and do my research at the
inspiring environment of the MPI Informatik, and Prof. Dr. Marc
Alexa, who ignited my interest in Computer Graphics in the first
place. I would also like to thank all the senior researchers here at
the institute who supervised me over the past years: Prof. Dr. Stefan
Gumhold,forguidingmethroughthebeginningofmyPhD;Dr. Zachi
Karni for motivating me to do research on Geometric Modeling; And
finally, Dr. Christian Theobalt for sparking my interest in motion
capture and all his support in the last few years. I am indebted to all
of them for their advice and guidance in my research.
Without the cooperation, advice and discussions of all my former and
present colleagues at the MPI many of my research projects would
not have been possible. I am especially grateful to all the co-authors
of the papers I have worked on during my PhD : Christian R¨ossl,
Hitoshi Yamauchi, Edislon de Aguiar, Naveed Ahmed, Nils Hasler,
Martin Sunkel, Bodo Rosenhahn, Thorsten Thorm¨ahlen and Ju¨rgen
Gall. Further, I would like to thank Boris Ajdin, Martin Fuchs and
Oliver Schall for the lively discussions, and Conny Liegl and Sabine
Budde for their help in managing day to day office life.
I would also like to thank our Maria Jacob, Yvonne Flory, Samir
Hammann, Lukas Ahrenberg and Tatjana Feldmann for allowing us
to record their performances and use them for research projects, and
Derek D. Chan for his help in dubbing our videos.
Finally,Iwouldliketothankmyfamilyfortheirsupport,andNatascha
for just being there for me all the time.
vviContents
1 Introduction 1
1.1 Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Contributions and structure . . . . . . . . . . . . . . . . . . . . . 4
1.3 List of publications . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2 Fundamentals 11
2.1 Basic data structures . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.1 3D objects and their representations . . . . . . . . . . . . 11
2.1.2 Images and videos . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Scanning and surface reconstruction . . . . . . . . . . . . . . . . . 15
2.2.1 3D scanning . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.2 Surface reconstruction . . . . . . . . . . . . . . . . . . . . 19
2.3 Shape editing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1 Linear methods . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.2 Non-linear methods . . . . . . . . . . . . . . . . . . . . . . 23
2.4 Physical simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4.1 Cloth simulation . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 Performance capture . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5.1 Motion capture . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5.2 3D video . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5.3 Performance capture . . . . . . . . . . . . . . . . . . . . . 29
I Differential coordinate based shape processing using
surfaces 31
3 A deformation framework for triangle mesh based templates 35
3.1 Differential representation . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Reconstruction and deformation . . . . . . . . . . . . . . . . . . . 39
viiCONTENTS
3.2.1 Constraint types . . . . . . . . . . . . . . . . . . . . . . . 41
3.2.2 Harmonic interpolation . . . . . . . . . . . . . . . . . . . . 42
3.2.3 Rotational invariance . . . . . . . . . . . . . . . . . . . . . 43
4 Inverse texture mapping 47
4.1 Initial deformation . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2 Surface Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5 Template based shape reconstruction 61
5.1 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2 Initial deformation and global scaling . . . . . . . . . . . . . . . . 65
5.3 Iterative improvement . . . . . . . . . . . . . . . . . . . . . . . . 66
5.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.5 Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.5.1 Laplacian updating . . . . . . . . . . . . . . . . . . . . . . 72
5.5.2 Remeshing . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.5.3 Surface fairing . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.5.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6 Surface based animation and performance capture 79
6.1 Data acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.2 Animation and tracking . . . . . . . . . . . . . . . . . . . . . . . 80
6.2.1 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.3 Model refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.3.1 Silhouette refinement using positional constraints . . . . . 84
6.3.2t using line constraints . . . . . . . . . 86
6.3.3 Multi-view stereo refinement . . . . . . . . . . . . . . . . . 86
6.3.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
II Differentialcoordinatebasedshapeprocessingusing
volumetric data 93
7 A deformation framework for tetrahedral meshes 97
7.1 Different