3D Reconstruction and Visualization of Alternatives for
Restoration of Historic Buildings
A New Approach
Lemonia Ragia
Froso Sarri
and Katerina Mania
School of Architectural Engineering, Technical University of Crete, Kounoupidiana, Chania, Greece
Department of Electronic and Computer Engineering, Technical University of Crete, Kounoupidiana, Chania, Greece
esarri@isc.tuc.gr, k.mania@ced.tuc.gr
Keywords: 3D Reconstruction, Geodetic Measurements, Computer Graphics.
Abstract: This paper puts forward a 3D reconstruction methodology applied to the restoration of historic buildings
taking advantage of the combined speed, range and accuracy of a total geodetic station. The measurements
of geo-referenced points produced a fully interactive and photorealistic geometric mesh of an historic
monument named ‘Neoria’. ‘Neoria’ is a Venetian building located by the old harbour at Chania, Crete,
Greece. The integration of tacheometry acquisition and computer graphics puts forward a novel integrated
software framework for the accurate 3D reconstruction of a historical building. The main technical
challenge of this work was the production of an accurate 3D mesh based on a sufficient number of
tacheometry measurements acquired fast and at low-cost. Interpolation methods ensured that a detailed
geometric mesh was constructed based on a few points. Advanced interactive functionalities are offered to
the user in relation to identifying restoration areas and visualizing the outcome of such works in a fully
interactive application based on game engine technologies. Moreover, the user could photorealistically
visualize the actual or restored monument and calculate distances between points.
During the last decade, the heritage community has
witnessed an increase in three-dimensional (3D)
photorealistic reconstructions of archaeological
structures which focused more on pleasing visuals
often overriding the question of scientific accuracy.
Assisted by advances in technology and cheaper
technical costs, museums, archival institutions and
heritage attractions have been frequently developing
digital content. However, archaeologists and
computer scientists have urged caution in the
abundant use of inaccurate and rushed 3D
reconstructions because of the possibility of
misleading the public. 3D reconstruction methods
should promote and ensure that scientific and
historical accuracy is driving such developments.
The ability to reconstruct a historical building’s
current or past structure is an invaluable tool which
could drive its potential restoration.
Restoration of historic buildings is conducted by
a team of historians, archaeologists and engineers
working together to provide restoration solutions
after the diagnosis and evaluation of each restoration
issue. 3D digitization of current historical structures
and 3D reconstruction of restoration areas should be
based on accurate on-site measurements (Sifniotis et
al., 2010). Tacheometry acquisition methods
(Boochs et al., 2006) use modern total stations that
result in high accuracy point measurements as well
as quick and efficient spatial data acquisition in
order to create a geometric mesh of existing
geometry structures. The aim of realistic image
synthesis and computer graphics techniques is the
creation of accurate, high quality imagery that
faithfully represent the geometric structure and light
propagation in a physical environment, the ultimate
goal being to create a 3D visualization which is
perceptually indistinguishable from an actual scene.
Reliable geometric measurements are necessary to
create a canvas of geo-referenced co-ordinates
which could be manipulated by expert users. In this
paper, the integration of tacheometry acquisition and
computer graphics puts forward a novel integrated
software framework for the scientifically accurate
3D reconstruction of an historical building in Crete,
Greece. Initially, a detailed 3D geometric model of
Ragia L., Sarri F. and Mania K..
3D Reconstruction and Visualization of Alternatives for Restoration of Historic Buildings - A New Approach.
DOI: 10.5220/0005376700940102
In Proceedings of the 1st International Conference on Geographical Information Systems Theory, Applications and Management (GISTAM-2015), pages
ISBN: 978-989-758-099-4
2015 SCITEPRESS (Science and Technology Publications, Lda.)
the monument is created based on acquired
terrestrial points on-site. Novel interpolation
methods ensure that a detailed geometric mesh is
constructed fast based on far fewer points available
compared to laser scanning. The geometric mesh is
lit utilizing modern global illumination algorithms.
Advanced interactive functionalities are offered to
the user based on game engine technologies. The
user can interactively manipulate as well as navigate
the geometric model. Based on adequate point
acquisition, the user can determine the position and
extent of erosions in the material surfaces, whether
the surface is smooth or includes bumps, whether
there are cleaved and damaged pieces or cracks and
the size of determined areas that subsequently
require full restoration. It can also be investigated
whether the edges of the walls and roof are vertically
aligned or whether there are related discrepancies.
The goal is to offer to the expert user an interactive
scientific tool which determines whether the
building needs maintenance and what interventions
need to be pursued.
1.1 Motivation
The building named ‘Neoria’ (Figure 1) is a
beautiful historic building located at the old harbor
of the old city of Chania, Greece. It was completed
in 1599 by the Venetians, consisting of seven
continuous domes and used for ship repair during
the winter time.
Figure 1: Neoria Historical Building, Chania, Crete.
The current state of this monument is
compromised as it stands a few meters from the
water facing, at times, adverse climate conditions.
As shown in Figure 2, parts of the monument’s stone
walls have fallen, entire sections of walls are loose
and certain edges of the building are partially or
entirely destroyed. Significant structural elements
are impaired and, as a result, Neoria is slowly losing
its original beauty. Cracks are now appearing and
restoration efforts had been limited or wrongly
conducted. For instance, in order to repair cracks,
craftsmen had often applied unsuitable modern
materials such as cement. The outcome of such
badly managed restorations was that at times, repairs
had destroyed neighboring elements to the
restoration areas rather than preserving them.
Figure 2: Damaged areas of Neoria in need of restoration.
The scope of this work is to provide
archaeologists with a technological and relatively
low-cost solution which will enable them to
reconstruct the monument at its current state fast.
Tacheometry measurements on-site are part of a
detailed dataset of the monuments facades and
interior. Such measurements could provide a
geometric mesh which enables expert users to
further inspect the digitally reconstructed monument
in detail and at every angle when planning its
restoration. Unlike time-consuming and lengthy
laser scanning processes, the methodology proposed
is fast, low-cost and of high scientific accuracy. In
this paper, we present the integrated software
framework based on game engine technologies for
the scientific 3D reconstruction and visualization of
Neoria in Chania, Crete employing a modern total
station and advanced computer graphic techniques.
1.2 Previous Work
Two 3D data acquisition methods have been widely
documented in research literature as part of the 3D
reconstruction process of archeological monuments:
photogrammetry and laser scanning.
Photogrammetry is a widely-adopted image-based
process requiring specialized hardware and software
such as a non-standard digital camera for
photogrammetry and a digital station (Gruen, 1998);
(Mayer et al., 2004); (Kersten, 2006); (El Hakim et
al., 2007); (Schaich, 2004). It is widely used for the
3D digitization of historical monuments (Lingua et
al., 2003). Although photogrammetry’s end result is
a high-resolution 3D model including geometric data
for restoration, it is a complex procedure that
requires specialized equipment and software. Dark
shadows and low luminance areas of structures
prevent efficient data acquisition resulting in
uncompleted geometry. On the other hand, laser
scanning technology extracts high resolution point
clouds and it is widely used for the scientific
documentation of historical monuments (Boehler et
al., 2004). 3D laser scanning allows the
reconstruction of individual parts of historical
monuments producing a fine geometry mesh (Allen
et al., 2003). Laser scanning supports the
maintenance, preservation and restoration of
monuments based on a highly detailed spatial data
set (Balzani et al., 2004); (Julia et al., 2010). Both
photogrammetry and laser scanning create an
accurate geometric mesh and provide rich detail.
However, in the process, both produce a large
amount of data and the post processing is quite
complicated (Grussenmeyer et al., 2008). In our
approach we use as an alternative geodetic total
stations which are portable, inexpensive, widely
available, requiring less training and having a wide
range of other applications. The goal of this
approach is to deploy a relatively low-cost total
station in order to acquire the co-ordinates of
specific surface points of Neoria in Chania’s harbour
area. The advantage of using a total station to
acquire measurements on-site is that it performs fast
digital data collection of spatial co-ordinates of
archaeological monuments. It is an optical
instrument used to measure horizontal and vertical
angles in order to determine relative position. The
main technical challenge was to produce an accurate
3D geometric mesh based on an efficient amount of
spatial co-ordinates that the total station captured in
the field. The end-goal was to implement an
interactive visualization framework based on game
engine technologies enabling experts to identify
areas for restoration.
1.3 Methodology
In this paper, we propose an optimized tacheometry
surveying technique employing a total station for the
monument surveying. The measurement technique
put forward is quick, flexible and accurate, requiring
a small number of points acquired in interest areas
which at the final stage of processing are connected
to form a geometric mesh (Andrews et al., 2009). If
needed, additional points can be easily acquired in
places and used in conjunction with previously
acquired datasets.
The 3D model is constructed by connecting the
acquired 3D points in order to reconstruct a
geometric mesh of a surface. Surface reconstruction
algorithms take as input a set of sample points that
describe the shape or topology of an object in three
dimensions and convert these points into a 3D
model. There is a vast variety of algorithms which
reconstruct the topology surface from sample points
in three dimensional space. The input of such
algorithms is a set of 3D coordinates utilized to
construct a polygonal mesh. A mesh consists of
vertices, edges and polygons. The 3D points are
considered vertices forming edges and polygons.
Reconstruction algorithms which compute
geometries either create the mesh using the existing
points as vertices or employ implicit functions to
approximate the surface. Algorithms utilizing
implicit surfaces documented in research literature
are the Poisson surface reconstruction (Kazhdan et
al., 2006) and Hoppe’s algorithm (Hoppe et al.,
1992). Popular algorithms which compute
geometries are the powercrust algorithm (Amenta et
al., 2001) and the ball pivoting algorithm
(Bernardini et al., 1999). After the surface is
reconstructed, subdivision is often needed.
Subdivision surfaces are polygon mesh surfaces
generated from a starting mesh through an iterative
process that levels the mesh while increasing its
density. Complex smooth surfaces can be derived in
a reasonably predictable way from relatively simple
meshes. There are two kinds of subdivision
schemes; approximating subdivision surfaces and
interpolating subdivision surfaces. Approximating
means that the result surfaces approximate the initial
meshes and that after subdivision, the newly
generated points are not included in the initial
surfaces. Examples of such subdivision algorithms
are Catmull–Clark (Catmull and Clark, 1978), and
Doo-Sabin subdivision surface (Doo and Sabin,
1978). Interpolating subdivision means that after the
subdivision, the control points of the original mesh
are interpolated to form the resulting surface. A
popular method of interpolating subdivision is the
butterfly subdivision (Zorin et al., 1996). Although
widely used when dealing with laser scanning point
clouds, these algorithms require dense point clouds
to produce an accurate model. The workflow of 3D
reconstruction based on a rather small amount of
acquired measurements utilizing a total station is
detailed in the following sections.
2.1 Data Acquisition
The outlines of the seven buildings of the Neoria
complex were initially measured, along with the
edges of the doors and the windows. Due to the
extent of the erosion, the framework proposed
provides measurements of the angles and distance to
surveyed points recorded digitally. Subsequently, the
x, y and z coordinates of each point were calculated
and connected, offering a 3D model of the
monument. A distance accuracy of ±2mm to ±5mm
was inherent to each point. In order to survey the
whole monument, the total station acquired
measurements based on selected positions called
‘stations’. The number of spatial points that were
measured was about 780.
2.2 Data Processing
The raw data were acquired in the form of the Greek
geodetic system coordinates. Civil3d by Autodesk,
processes geodetic data in the form of coordinates as
input and converts them to 3D points. The raw data
were imported as Greek Geodetic Reference System
GGRS 87 coordinates. A survey database was
created and the points were treated as point groups,
processed concurrently or in parts as needed. The
coordinates acquired are now treated as 3D points in
space, forming a sparse 3D point cloud. The main
technical challenge of this phase was to produce an
accurate 3D mesh based on a low resolution 3D
point cloud. The spatial data acquisition method
adopted, although quick and efficient, it doesn’t
provide a large volume of information in the form of
a high resolution point cloud that other methods
such as laser scanning achieve. We will explain, in
the following sections, how an accurate geometric
mesh was produced based on a sufficient number of
geodetic measurements rather than on a dense, often
complex, point cloud.
2.3 Surface Reconstruction
Surface reconstruction algorithms vary depending on
how they handle available spatial data (Tang et al.,
2013). In order to reconstruct a 3D surface based on
total station measurements, approximating
algorithms were quickly dismissed. These
algorithms require dense point clouds to produce an
accurate model. In this case here, the exact geodetic
measurements acquired in the field were to be
interpolated in order to produce a finer geometric
mesh. An appropriate interpolation algorithm would
take as input a sufficient amount of points recorded
via the geodetic total station and identify spatial
areas where interpolation is needed in order to fill in
gaps appearing in-between measurements. An
example of the implementation of the ball pivoting
algorithm as applied to a small subset of the total
station measurements is shown in Figure 3. There
are holes and geometry discontinuities due to lack of
surface information and varying density of the data.
The best approach would be to use Delaunay
triangulation to form a Triangular Irregular Network
(TIN surface) (Lee and Arthrur, 1986);
(Edelsbrunner, 2000).
Figure 3: Implementation of the ball-pivoting algorithm.
Holes and discontinuities are visible.
2.4 Delaunay Triangulation
Given a set of data points, Delaunay triangulation
produces a set of lines connecting each point to its
natural neighbors. The Triangular Irregular
Networks are used to represent surfaces in
Geographic Information Systems (GIS - earth
surfaces). In order to produce surfaces consisting of
the four walls of the building, the axis of data was
rotated 90 degrees facing upwards. The points
belonging to each wall were processed separately.
The x, y, z dimensions of each wall were designated
appropriately to achieve the desired rotation. After
implementing the necessary changes, four separate
groups of data were derived subsequently processed
as TIN surfaces. The points were connected to form
triangles according to the Delaunay methodology.
These networks were used as base surfaces (Figure
4), however, further processing was necessary in
order to add further surface detail so that the real-
world surface is approximated.
Figure 4: Delaunay Triangulation of the survey points.
2.5 Surface Processing
Initially, each wall was rotated so that they were
correctly aligned attached to each other, forming the
complete mesh of the monument. Depth information
was integrated around the monument’s windows and
doors and the mesh was ‘tweaked’ wherever needed
in order to correct limited mesh discontinuities. Each
side of the building was processed and then
subsequently aligned to form the complete building
(Figure 5, 6).
At this point, a basic shape of the outline of the
monument including details of its structure was
completed. In order to integrate further detail, the
surface was further subdivided. It was significant to
maintain the original position of the vertices so that
an interpolating subdivision could be deployed to fill
in the gaps.
Figure 5: The base surface after corrections (South side).
The depth of the doors and windows is visible.
Figure 6: The North side of Neoria after geometry
The next step was to take into consideration the
shape and quality of the surface. Large segments of
the monument included insufficient spatial
information because of the small amount of points
recorded. For example, there were sparse blank
spaces between windows. Additional spatial
information around these areas was paramount in an
effort to maintain the monument’s initial shape. It
was evident that interpolating subdivision methods
deform such areas. By not succeeding to find surface
information, they randomly add points in blank
spaces, deforming the surface and negatively
affecting the geometric fidelity of the model. A
simple midpoint subdivision was applied to the
spatial dataset (Chen and Prautzsch 2012). The
subdivision scheme was applied at the point where
every edge was split on its midpoint. A finer mesh
with no distortion was acquired at the end of this
process (Figure 7, 8).
To further approximate the real-world surface,
the subdivided mesh was refined. A finer and
smoother surface enables the application of detailed
texturing on the surface making it more realistic.
The problematic areas such as the circular windows
and doors were initially brought into focus. To
accomplice a realistic circular window, the edges of
the windows were further subdivided by applying a
Meshsmooth modifier in a standard modeling
software, which subdivides the geometry while
interpolating the angles of new faces at corners and
edges. This process resulted in the desired geometric
shape around the problematic areas (Figure 8).
Figure 7: The 3D model after implementing midpoint
Figure 8: Window details after further refinement.
2.6 Texturing
The 3D surface was textured applying images
acquired on-site. In order to begin the process of
texturing, the current, largely un-restored condition
of the monument was taken into account. There are
cracks, damages and alterations on the walls. Parts
of the brick walls are missing and there are add-ons
of modern materials applied to certain parts. The
challenge was to apply such complex effects on the
available plane polygonal faces. For this purpose,
multi-texturing was employed. The primary set of
textures included mostly images taken on the field.
Normal mapping was applied which simulated the
bumps and abnormalities of the surface by detecting
the edges of the surface’s constructed elements such
as bricks, break lines, etc. offering the illusion of an
uneven surface.
UV mapping is the process of forming a 2D
image representation out of a 3D model, in order to
correctly project a texture map on the model. During
this process, the model’s derived 3D mesh was
unwra90pped to a flat 2D image. This process
involved splitting the geometric mesh into clusters
by selecting similarly-looking parts and faces. In
each cluster, the appropriate texture was applied as a
diffuse map. The diffuse map is a texture used to
define the surface's main colour. It sets the tint and
intensity of diffuse light reflection on the surface. A
simple diffuse map, however, does not communicate
photorealism. The perception of realism as well as
3D depth is attained by normal mapping. Normal
mapping simulates small displacements of the
surface while the surface geometry is not modified.
Instead, only the surface normal is modified as if the
surface had been displaced. This is achieved by
perturbing the surface normal of each object and
using the perturbed normal during lighting
Normal maps consist of red, green, and blue.
These RGB values translate to x, y, and z
coordinates, allowing a 2D image to represent depth.
The 2D image is applied to the surface. This way, a
3D surface simulates the lighting and ultimately the
color associated with 3D coordinates. In the normal
map, each pixel's color value represents the angle of
the surface normal.
After texturing, the model was imported in the
Unity3D game engine. Game engines have become a
tremendous asset in archaeological and architectural
visualization. Unity3D is a powerful cross-platform
game engine developed by Unity Technologies. It is
used not only to develop video games, but also for
any kind of sophisticated 3D visualization which can
be ported to the web or on any mobile platform.
2.7 Lighting
Photorealism relies heavily on lighting. After
texturing, it is essential to apply a global
illumination algorithm so that the lighting of the
scene simulates real-world lighting conditions.
Unity3D uses the Beast lighting algorithm. Unity
Pro extends this functionality by Global
Illumination, allowing for the so-called ‘baking’ of
realistic and beautiful lighting, that would otherwise
be impossible in real-time.
Global illumination represents a group of
algorithms used in 3D computer graphics offering
physics-based rendering by emulating inter-
reflections in a scene. Such algorithms take into
account not only the light which comes directly from
a light source (direct illumination, local
illumination), but also perceives all surfaces in the
scene as lights sources which, in turn, affect the
lighting of other surfaces, whether reflective or not
(indirect illumination). In fact, this is a recursive
process which should be terminated when the global
illumination algorithms are utilized. It is user-
defined when the algorithm will terminate; in most
cases this occurs when subsequent iterations do not
produce perceivable additional lighting effects and
shadowing compared to the previous ones. Diffuse
inter-reflection is a global illumination effect applied
to architectural visualization and to the interior or
exterior of buildings producing beautiful subtle
shadowing and color-bleeding effects, integrated in
the radiosity global illumination algorithm. The
Beast lighting algorithm is a radiosity-like renderer
that enhances the photorealism of the geometric
mesh and offers a realistic as well as scientifically
accurate simulation of the Neoria monument
(Figures 9, 10).
Figure 9: The North side of the Neoria monument
photorealistically rendered.
Figure 10: The South side of the Neoria monument
photorealistically rendered.
The goal of this work was to develop an interactive
visualization tool which offers expert users the
capability of interacting with a scientific 3D
reconstruction of an existing monument acquiring
information concerning its future restoration. The
visual impact of such restoration is rendered after
user interaction and presented to the user. Unity3D
offers sophisticated tools for the development of 3D
interactive applications and a programming interface
utilized for the development of complex geometry
behaviors, user interaction, etc. The final
application’s functionalities were programmed in
3.1 3D Interactive Visualization
Interactive functionalities are provided to the user
who dynamically manipulates as well as navigates
the geometric model (Figure 11). The expert user
can zoom on parts of the geometry for closer
inspection and identify interest areas relevant to
monument maintenance, erosion, wall alignment,
cracks, measurements, etc.
Figure 11: Interactive walkthrough of the monument and
Figure 12: Currently, there is a missing part on top of the
Advanced functionalities allow the user to alter
the geometry provided. This operation is
implemented in damaged areas which are in need of
restoration represented as geometric mesh (Figure
12, 13). For instance, added cement can be removed
and replaced visualized by acquiring textures
derived from non-damaged areas. Hotspots of
interest and restoration areas can be activated
through the on-screen menu. By interacting with the
menu, the user can toggle between surfaces; for
example, the user can visualize the monument
surface as it exists today and the restored surface
rendered in red color (Figures 13, 14). In order to
render the restored surface, the appearance of the
non-damaged parts was taken into consideration, as
well as the measurements of structural elements of
neighbouring areas of interest. This process
recreated missing surfaces while there was lack of
architectural information.
Figure 13: Digital restoration of the missing part; the red
part does not exist at present.
Figure 14: Currently, the door opening is blocked.
Figure 15: Door opening restored as appearing in the past.
Scientific information is offered to the expert
user in relation not only to potentially restored areas
but also to measurements between areas, existing
and restored. The user could select specific points on
the surface and the system calculates the distance
between them. Moreover, the user could actually
visualize the co-ordinates of the original tachometry
measurements and identify areas where the points
are dense or scarce (Figure 16). The user could also
visualize the current condition of the monument lit
under different lighting conditions. In order to aid
further the specialists in restoration to extract
architectural information, a functionality which can
display information was implemented. This
operation can be chosen from the screen menu,
prompting the user to pick two points anywhere on
the model and information regarding the position
and relation of the chosen points is displayed. Their
distance in 3d space is calculated, as long as their
depth difference on the vertical surface (Figure 17).
Figure 16: Display of original tacheometry measurements
(green points) and their coordinates.
Figure 17: Distance calculated between points. The upper
corners of a window were chosen for measurement.
The units are in meters. In that way each user can
measure the dimensions of a particular feature or can
gain information on the size and exact place of a
restored part.
In this paper, the integration of tacheometry
acquisition and computer graphics puts forward a
novel integrated software framework for the
scientifically accurate 3D reconstruction of a
historical building in Crete, Greece. The interactive
framework presented was based on game engine
technologies which provide powerful tools for
photorealistic visualization. The main technical
challenge of this work was the production of a
scientifically accurate 3D mesh based on a rather
small number of tacheometry measurements
acquired fast and at low-cost. Novel interpolation
methods ensured that a detailed geometric mesh was
constructed fast based on far fewer points available
compared to laser scanning. Advanced interactive
functionalities are offered to the user in relation to
identifying restoration areas and visualizing the
outcome of such works. Moreover, the user could
visualize the co-ordinates of the points measured,
calculate distances at will and navigate the complete
3D mesh of the monument. Future work will
produce an advanced visualization system operated
on a mobile platform, investigating the effectiveness
of complex visualization techniques linked to
diverse restoration strategies by local authorities,
visualizing the visual impact of such restorations.
We would like to thank Anastasia Tzigounaki,
serving as the director of the 15th Ephorate of
Prehistoric and Classical Antiquities in Chania,
Greece, for her contribution to this work.
Andrews, D, Bedford, J, Blake, B and Bryan, P., 2009.
Metric Survey Specifications for Cultural Heritage, 2
edition. Swindon: English Heritage.
Allen, P., K., Troccoli, A., Smith, B., Murray, S., Stamos,
I., and Leordeanu, M., 2003. New methods for digital
modeling of historic sites. IEEE Computer Graphics
and Applications 23, no. 6: 32-41.
Amenta, N., Choi, S., & Kolluri, R. K., 2001. The power
crust. In Proceedings of the sixth ACM symposium on
Solid modeling and applications, pp. 249-266.
Balzani, M., Santopuoli, N., Grieco, A., & Zaltron, N.
2004. Laser scanner 3D survey in archaeological field:
The Forum of Pompeii. In International Conference
on Remote Sensing Archaeology, Beijing, pp. 169-175.
Bernardini, F., Mittleman, J., Rushmeier, H., Silva, C., &
Taubin, G., 1999. The ball-pivoting algorithm for
surface reconstruction. Visualization and Computer
Graphics, IEEE Transactions on, 5(4), 349-359.
Boehler, W., Vicent, M. B., Heinz, G., Marbs, A., &
Müller, H., 2004. High quality scanning and modeling
of monuments and artifacts. InProceedings of the FIG
Working Week, pp. 22-27.
Boochs, F., Hoffmann, A., Huxhagen, U., and D. Welter,
D., 2006. Digital reconstruction of archaeological
objects using hybrid sensing techniques-the example
1568: 395.
Catmull E. and Clark J., 1978. Recursively Generated B-
Spline Surfaces on Arbitrary Topological Meshes,
Computer Aided Design, vol. 10, no. 6, pp. 350-355.
Chen, Q., and Prautzsch, H., 2012. General Midpoint
Subdivision. arXiv preprint arXiv:1208.3794.
Doo, D., and Sabin, M., 1978. Behaviour of recursive
division surfaces near extraordinary points. Computer-
Aided Design, 10(6), 356-360.
Edelsbrunner, H., 2000. Triangulations and meshes in
computational geometry. Acta Numerica 9, 133-213.
El-Hakim, Gonzo L., Voltolini F., Girardi S., Rizzi R.,
Remondino F., Whiting E., 2007. Detailed 3D
Modelling of Castles. International Journal of
Architectural Computing, Vol. 5(2), pp. 199-220.
Gruen, A., 1989. Digital photogrammetric processing
systems- Current status and prospects.
In (International Society for Photogrammetry and
Remote Sensing, Congress, 16 th, Kyoto, Japan, July
1-10, 1988) Photogrammetric Engineering and
Remote Sensing, Vol. 55, pp. 581-586.
Grussenmeyer, P., Landes, T., Voegtle, T.,and Ringle, K.,
2008. Comparison methods of terrestrial laser
scanning, photogrammetry and tacheometry data for
recording of cultural heritage buildings. In ISPRS
Congress Proceedings, Beijing, pp. 213-18.
Hoppe, H., DeRose, T., Duchamp, T., McDonald, J., &
Stuetzle, W., 1992. Surface reconstruction from
unorganized points Vol. 26, No. 2, pp. 71-78.
Julia, A., G., Belén R., R., Diego G., A., and M. Teresa R.,
B., 2010. Terrestrial laser scanning intensity data
applied to damage detection for historical
buildings. Journal of Archaeological Science37, no.
12: 3037-3047.
Kazhdan, M., Bolitho, M., & Hoppe, H., 2006. Poisson
surface reconstruction. In Proceedings of the fourth
Eurographics symposium on Geometry processing.
Kersten, T., 2006. Combination and Comparison of
Digital Photogrammetry and Terrestrial Laser
Scanning for the Generation of Virtual Models in
Cultural Heritage Applications. The 7th International
Symposium on Virtual Reality, Archaeology and
Cultural Heritage, VAST. pp. 207–214, M. Ioannides,
D. Arnold, F. Niccolucci, K. Mania (Editors).
Lee, D., T., and Arthur K., L., 1986. Generalized
Delaunay triangulation for planar graphs. Discrete &
Computational Geometry 1.1 201-217.
Lingua, A., M., Piumatti, P., and Rinaudo, F., 2003.
Digital photogrammetry: a standard approach to
cultural heritage survey. International Archives of the
Photogrammetry, Remote Sensing and Spatial
Information Sciences 34.Part 5: 210-215.
Mayer H., Mosch M., Peipe J., 2004. 3D model generation
and visualization of Wartburg Castle ISPRS Int.
Workshop on Processing and Visualizaation Using
High-Resolution Imagery. Vol (36) PART 5/W1, 18-
20 November, Pitsanulok Thailand.
Schaich, M., 2004. 3D Scanning Technologies and Data
Evaluation in an Archaeological Information
System. Proceedings of the CAA. 2004.
Sifniotis, M., Jackson, B., Mania, K., Watten, P.L., White,
M., 2010. 3D Visualization of Archaeological
Uncertainty. Poster, ACM Siggraph Symposium on
Applied Perception in Graphics and Visualization
2010, Los Angeles, USA.
Tang R., Halim S., Zulepli M., H., 2013. Surface
Reconstruction Algorithms Review and Comparison.
Zorin, D., Schroder, P., and Sweldens, W., 1996.
Interpolating subdivision for meshes with arbitrary
topology. In Proceedings of the 23rd annual
conference on Computer graphics and interactive
techniques (pp. 189-192) ACM.