Generation of a 3D city model of Baalbek, Lebanon based on historical photos [Elektronische Ressource] / vorgelegt von Ahmed Alamouri

Generation of a 3D city model of Baalbek, Lebanon based on historical photos [Elektronische Ressource] / vorgelegt von Ahmed Alamouri

-

Documents
225 pages
Obtenez un accès à la bibliothèque pour le consulter en ligne
En savoir plus

Description

   Generation of a 3D city model of Baalbek/Lebanon based on historical photos  Vorgelegt von Dipl.-Ing. AHMED ALAMOURI Von der Fakultät VI-Planen Bauen Umwelt der Technischen Universität Berlin Zur Erlangung des akademischen Grades Doktor der Ingenieurwissenschaft (Dr.-Ing.) genehmigte Dissertation Promotionsausschuss: Vorsitzender: Univ. Prof. Dr. Ing. DOROTHÉE SACK Berichter: Prof. Dr. Ing. Dr. h. c. LOTHAR GRÜNDIG Berichter: Prof. Dr. rer. nat. THOMAS H. KOLBE Berichter: Prof. Dr. Ing. ORHAN ALTAN Tag der wissenschaftlichen Aussprache: 27.01.2011 Berlin 2011 D83 ii PhD-Thesis by Alamouri   Generation of a 3D city model of Baalbek/Lebanon based on historical photos  Promotionsausschuss: Vorsitzender: Univ. Prof. Dr. Ing. DOROTHÉE SACK Berichter: Prof. Dr. Ing. Dr. h. c. LOTHAR GRÜNDIG Berichter: Prof. Dr. rer. nat. THOMAS H. KOLBE Berichter: Prof. Dr. Ing. ORHAN ALTAN Tag der wissenschaftlichen Aussprache: 27.01.2011 Berlin 2011 D83   PhD-Dissertation by Alamouri iii   iv PhD-Thesis by Alamouri ACKNOWLEDGEMENTS At the outset, I would like to express my profound and gratitude to my supervisor Prof. Dr. Eng. Dr. h. c. Lothar Gründig for his supports and advices during my doctoral research. He has constantly kept me to remain focused on achieving the research aims.

Sujets

Informations

Publié par
Ajouté le 01 janvier 2011
Nombre de lectures 59
Langue English
Signaler un abus

 
 
 
Generation of a 3D city model of 
Baalbek/Lebanon based on historical photos 


Vorgelegt von
Dipl.-Ing.
AHMED ALAMOURI

Von der Fakultät VI-Planen Bauen Umwelt
der Technischen Universität Berlin
Zur Erlangung des akademischen Grades
Doktor der Ingenieurwissenschaft (Dr.-Ing.)
genehmigte Dissertation

Promotionsausschuss:
Vorsitzender: Univ. Prof. Dr. Ing. DOROTHÉE SACK
Berichter: Prof. Dr. Ing. Dr. h. c. LOTHAR GRÜNDIG
Berichter: Prof. Dr. rer. nat. THOMAS H. KOLBE
Berichter: Prof. Dr. Ing. ORHAN ALTAN

Tag der wissenschaftlichen Aussprache: 27.01.2011
Berlin 2011
D83 ii PhD-Thesis by Alamouri
 
 
Generation of a 3D city model of 
Baalbek/Lebanon based on historical photos 







Promotionsausschuss:
Vorsitzender: Univ. Prof. Dr. Ing. DOROTHÉE SACK
Berichter: Prof. Dr. Ing. Dr. h. c. LOTHAR GRÜNDIG
Berichter: Prof. Dr. rer. nat. THOMAS H. KOLBE
Berichter: Prof. Dr. Ing. ORHAN ALTAN

Tag der wissenschaftlichen Aussprache: 27.01.2011
Berlin 2011
D83 
 
 PhD-Dissertation by Alamouri iii







 
 iv PhD-Thesis by Alamouri
ACKNOWLEDGEMENTS
At the outset, I would like to express my profound and gratitude to my supervisor Prof. Dr.
Eng. Dr. h. c. Lothar Gründig for his supports and advices during my doctoral research. He
has constantly kept me to remain focused on achieving the research aims. His observations
and comments helped me to establish the overall direction of the research and to move
forward with the investigation in depth. Giving me the opportunities to participate in many
workshops, meetings and conferences, he has raised my self-confidence and experience of
making presentations at scientific meeting.
As well, I would like to express my great thanks to Prof. Dr. rer. nat. Thomas H. Kolbe who
assisted me during the past five years. Very special thanks for his constant interest and
encouragement; which helped me complete this thesis.
Furthermore, my deep appreciation also expressed to Dipl. Inform. Gerhard König who was
helping to solve and overcome the problems arisen during the research work. In addition,
many thanks expressed to all colleagues at the Department of Geodesy and Geoinformation at
the Technical University of Berlin, Germany.
I am grateful also to Mr. Dipl.-Eng. Frank Henze (BTU-Cottbus) and to colleagues at Institute
of Francais du Proche Orient (IFPO) in Damascus for providing the materials used in this
research. In addition, thanks to University of Aleppo, Syria for the finance support during my
PhD program in Germany.
Last but not least, I am greatly indebted to my parents who supported me mentally and
patiently during my research time in Berlin. Thanks are extended to my family and friends for
sharing with me the happy moments and opinions about the life while in Germany.




Ahmed Alamouri
2011, Berlin - Germany
 
 PhD-Dissertation by Alamouri v
PREFACE
The work presented by this thesis was funded by the coorporation between the University of
Aleppo (Syria), Technical University of Berlin (Germany) and Brandenburg Technical
University of Cottbus (Germany) in the research area “the evaluation of digital
photogrammetry in the documentation of the cultural heritage”.
The investigation and the implementation of experiments are based on the data of the
historical graphical materials of the city Baalbek in Lebanon. This work was presented from
October 2005 until September 2010 at the Berlin Institute of Technology, University of Berlin
and has been supervised by Prof. Dr. Eng. Dr. h. c. Lothar Gruendig.
With this thesis, three main problems will be posed and investigated:
 The first, how can the orientation process of different historical image types be best
achieved (with respect to the poor properties of the images used)?
 Second, the combination between the different historical image types of Baalbek to achieve
an optimal 3D object reconstruction will be discussed.
 Finally, it will be investigated: in which Level of Detail (LOD) Baalbek’s data set can be
modelled with respect to requirements of CityGML (City Geography Markup Language).
The thesis comprises eight Chapters; the first one includes an introduction about the research
motivation and objectives. Some theory and mathematical background of camera calibration
ndwill be presented in the 2 Chapter. In the third one a new approach for relative orientation of
non-calibrated historical photos of Baalbek will be described. The fourth and fifth Chapters
describe the orientation process based on the combination between different image types. 3D
object acquisition, quality assessment of Baalbek’s data sets as well as a 3D modelling will be
th thdiscussed in the 6 and 7 Chapters.
thIn 8 Chapter, concluding remarks and results are summarized. In addition, perspectives,
remaining open questions and problems for future works are addressed.


 
 vi PhD-Thesis by Alamouri
ZUSAMMENFASSUNG
Diese Arbeit befasst sich mit der Auswertung von historischen Bildern unterschiedlichster Art
zur Dokumentation von Objekten des Weltkulturerbes. Beispielhaft wurde verfügbares
Bildmaterial (vertikale, terrestrische und Schrägaufnahmen) der Stadt Baalbek in Lebanon
verwendet und photogrammetrisch ausgewertet. Bei der Verarbeitung des historischen
Bildmaterials tritt eine Vielzahl von Problemen auf, deren Ursache in der Aufnahmesituation
(wie zum Beispiel: unterschiedliche Aufnahmezeitpunkte, unterschiedliche Bildmaßstäbe,
verschiedene Flughöhen und verschiedene Kameras) bzw. in der Beschaffenheit der Bilder zu
sehen ist (unterschiedlicher Kontrast der Grauwerte in den Bildern, Bildrauschen). In der
Regel sind keine Daten über die verwendeten Kameras bekannt, was eine Auswertung
wesentlich erschwert.
In dieser Arbeit wird eine Strategie entwickelt, die einen Beitrag zur kombinierten
Auswertung von historischen Bildern liefert. Der Lösungsansatz sieht drei Stufen vor:
 Unter diesen Bedingungen von den verwendeten Bildern wird in dieser Arbeit eine
entsprechende Strategie basiert auf verschiedenen Klassen von historischen Bildern zur
Kamerakalibrierung und Bildorientierung entwickelt. Diese Strategie liefert einen effektiven
Beitrag für den relativen Orientierungsprozess, da diese kombinierte Auswertung von den
historischen Bildern bisher nur unzureichend untersucht wurde. Sie bildet daher einen
Schwerpunkt dieser Arbeit.
Zunächst müssen die verfügbaren Bilder im Koordinatensystem Baalbeks gemeinsam
orientiert werden, um die unbekannten Parameter der inneren sowie der äußeren Orientierung
zu bestimmen. Diese Parameter werden durch einen speziellen Algorithmus auf Basis der
Bündelblockausgleichung berechnet und somit die Verzerrungsparameter (radial und
dezentral) ermittelt. In der Arbeit wird untersucht, wie die Näherungswerte der Unbekannten
optimal geschätzt werden und wie weit sie die erforderliche Genauigkeit und Zuverlässigkeit
beeinflussen.
 Zur relativen Orientierung der historischen Bilder von Baalbek wurde ebenfalls die
Methode der Bündelblockausgleichung angewendet, da sie den Zusammenhang zwischen den
Bildkoordinaten (als Beobachtungen) und den Objektkoordinaten direkt, d.h. ohne Umweg
über Modellkoordinaten, herstellen kann. Dieser Zusammenhang lässt sich durch die
 
 PhD-Dissertation by Alamouri vii
mathematische Beziehung zwischen den Beobachtungen (z. B. Bildkoordinaten,
Passpunkten, usw.) und den Unbekannten (wie z. B. die Parameter der inneren und äußeren
Orientierung) präsentieren. Diese Beziehung kann mathematisch als funktionales Modell
dargestellt werden, welches als die im Ausgleichungsprozess angewendete Hauptform gilt.
Generell ist der funktionale Ansatz nichtlinear, deshalb wird im Rahmen der Arbeit der
Lösung dieses Problems spezielle Aufmerksamkeit gewidmet. Unter der Annahme, dass die
Ausgleichung generell mit linearen Funktionen gelöst werden kann, wird ein Linearisierungs-
prozess eingeführt.
 In den jetzt orientierten Bildern können räumliche Objektkoordinaten gemessen werden,
um 3D-Objekte zu rekonstruieren. In dieser Arbeit wird untersucht, in wie weit durch die
Kombination der verschiedenen historischen Bilder eine beste 3D-Objektrekonstruktion
möglich ist.
Die 3D-Daten Baalbeks liefern den Input zu einer 3D-Modellierung basierend auf CityGML
(City Geography Markup Language). Daher beleuchtet die Arbeit, welcher Detaillierungsgrad
(Level of Detail-LOD) bei der Modellierung erreicht werden kann.
Darüber hinaus werden, soweit möglich, CityGML-Objekte (CityGML-Entitäten) der Stadt
Baalbek durch semantische Informationen angereichert (z. B. Kategorie, Name und Adresse
des Objekts), da sie einen wesentlichen Mehrwert für die Analyse der historischen
Entwicklung der Stadt liefern.
Mit der vorliegenden Arbeit wurde erreicht, dass die interne geometrische Konfiguration der
unterschiedlichen verwendeten Kameras unter Berücksichtigung der Linsenverzerrung
bestimmt werden konnte, und somit eine integrierte dreidimensionale Vermessung auf der
Basis historischer Bilder möglich ist. Als weiteres anwendungsbezogenes Ergebnis wurde ein
3D-Stadtmodell von der historischen Stadt auf der Basis von CityGML in zwei verschiedenen
LODs generiert.




 
 
.:????viii PhD-Thesis by Alamouri
ABSTRACT
Within this research the combined evaluation of different types of historical photos for the
documentation of cultural heritage sites is discussed. For this purpose the historical photos of
Baalbek in Lebanon have been used which can be classified into three types: vertical, oblique
and terrestrial photos. Due to the poor properties of these photos (such as: different cameras
used and no primary data of cameras’ parameters, different image scales, different altitudes of
the flight, the contrast of the gray values and the image noise are relative high, in addition the
images had been taken in different dates, etc.), it is not a priori clear by which combination of
the image orientation process the best results (namely, the accurate determination of camera
parameters) could be achieved. Therefore, an approach for the relative orientation process
will be discussed. Since the relative orientation is the position recovery and orientation of one
image system relative to another, the potential of the mentioned approach is to reveal (from a
mathematical point of view) the direct and indirect impact between the historical images of
Baalbek to each other. It can be considered as a new contribution for the image orientation
process based on the combination of different image types. This approach consists of three
steps; the first one is the orientation of each image type in a separated block with its special
properties (e.g. for each type a special camera was assumed). Second, the orientation process
only for the vertical and oblique photos together has been achieved. Finally, all types of
historic Baalbek’s photos have been assembled into the same block and then oriented.
The abovementioned orientation steps were carried out using the bundle block adjustment
method. This way has been selected because it enables to model the direct relation between
the photo’s reference system and object space optimally. The mentioned relation between
both systems describes the relationship between the input data which are called observations
(e.g. the coordinates of image points, control points, etc.) and unknown parameters (e.g.
the parameters of interior and exterior orientation). The requested relationship can be
mathematically presented through a model (called a functional model) which will be the main
form applied in the adjustment problem. In general, the functional model is highly non-linear;
therefore a special emphasis has to be respected when solving this problem. Due to the
assumption that the adjustment method is generally achieved with linear functions, a
linearization process should be enforced.
 
 
.:????PhD-Dissertation by Alamouri ix
Based on the oriented model of the historical photos, the combination between the different
historical images of Baalbek - in order to achieve a best possible 3D object reconstruction
will be evaluated by independent controls. Moreover, Baalbek’s 3D data extracted from
oriented historical photos will be considered the main input data used for a 3D CityGML
modelling (City Geography Markup Language). CityGML modelling supports different
Levels of Detail (LODs); by them different data collections can be represented and modelled.
In this context, it will be investigated: in which LOD Baalbek data acquired can be modelled!
In other words: based on the quality check of Baalbek data we want to know in which LOD
Baalbek data available could be modelled.
In addition, the semantic modelling of this city should be taken into account, because the
geometric model does not support all information needed (e.g. classification of objects,
building function, building class, names, etc.). Thus, the objects’ geometry isn’t the only
quality concern. This will guarantee the creation of a geometric and semantic 3D city model
of Baalbek which on the one hand provides an important document for this city and on the
other hand allows to understand and to enable the analysis of the historical developments of
thBaalbek’s building remains from the prehistoric date until 20 century.
Work results are: the internal geometric configuration of the different cameras used and the
lens systems were determined; so that the measurement of the applicable 3D object points is
possible. Moreover, the quality of Baalbek’s data extracted based on the oriented model of the
photos has been checked. Depending on the quality assessment of Baalbek’s data a 3D city
model of this historic city was created in two different LODs. The results will be also
considered a data base for other applications and projects in Baalbek’s space (for e.g. the
Geoinformation System GIS, archaeology, architecture, etc.).


 
 x Content
Content
Acknowledgements .................................................................................................................. iv
Preface ....................................................................................................................................... v
Abstract .................................................................................................................................. viii
1. Introduction .......................................................................................................................... 1
1.1. Motivations of the research ......................................................................................... 1
1.2. Research objectives ..................................................................................................... 7
1.3. Related work ............................................................................................................... 8
1.4. Available materials ................................................................................................... 10
1.4.1. Vertical aerial photos 10
1.4.2. Oblique aerial images ......................................................................................... 12
1.4.3. Terrestrial photographs ...................................................................................... 13
1.4.4. Additional information ....................................................................................... 13
2. Theory of a camera calibration 16
2.1. Interior orientation .................................................................................................... 16
2.2. Exterior orientation ................................................................................................... 20
2.2.1. Standard case ...................................................................................................... 20
2.2.2. Special terrestrial photogrammetry .................................................................... 22
2.3. Bundle Block Adjustment (BBA) ............................................................................. 24
3. A new approach for relative orientation of non-calibrated historical photos .............. 27
3.1. Overview of the orientation process ......................................................................... 27
3.2. Orientation of Baalbek’s historical vertical aerial images ........................................ 28
3.2.1. A primary preparation of vertical photos ........................................................... 28
3.2.1.1. Camera’s focal length .................................................................................. 28
3.2.1.2. Image scale determination ........................................................................... 29
3.2.1.3. Scan and ground resolution ......................................................................... 31
3.2.1.4. Determination of fiducial points .................................................................. 31