The document discusses the use of open-source tools and techniques for three-dimensional documentation in archaeology, emphasizing the capabilities of the Python Photogrammetry Toolbox (PPT) and Meshlab for creating 3D models from images. It highlights the advantages of these methods, including cost-effectiveness, ease of use, and the ability to achieve high accuracy in documenting archaeological sites. Additionally, the document presents various case studies showcasing the application of these techniques in real-world archaeological settings.

1. AIUM 2012 «ARCHEOLOGY INTERNATIONAL UNIVERSITY MEETING» 8th-10th NOVEMBER 2012
An Open Source solution for
Three-Dimensional documentation:
archaeological applications
Giulio Bigliardi
CGT-Centro di GeoTecnologie (Univ. di Siena)
bigliardi2@unisi.it
Marta Bottacchi
CGT-Centro di GeoTecnologie (Univ. di Siena)
bottacchi@unisi.it
Sara Cappelli
CGT-Centro di GeoTecnologie (Univ. di Siena)
cappelli11@unisi.it
Leonardo Carmignani
Dipartimento di Archeologia e Storia delle Arti –
Sezione di Preistoria e Protostoria (Univ. di Siena)
leocarmignani@msn.com

2. Digital models are nowadays present everywhere,
their use and diffusion are becoming very popular
through the Internet and they can be displayed
on low-cost computers.
Although creating a simple 3d model seems to be
quite simple, actually the generation of a precise
and photo-realistic computer model of a complex
object still requires considerable effort.
Three-dimensional digital models are required in
many applications such as inspection, navigation,
object identification, visualisation and animation.

3. Recently it has become a very
important and fundamental step
especially for cultural heritage digital
archiving. The aims are different:
documentation in case of loss or
damage, virtual tourism and
museum, education resources,
interaction without risk of damage,
and so forth.
The specific requirements for many
applications, including digital
archiving and mapping, involve high
geometric accuracy, photo-realism of
the results and the modelling of the
complete details, as well as the
automation, low cost, portability and
flexibility of the modelling technique.

4. Three-dimensional modelling of objects
and scenes is an intensive and long-
lasting research problem in the graphic,
vision and photogrammetric
communities.
3D Digital copy can be done by different
technology, laser (ground), lidar
(aerial), photogrammetry. Lidar
Nowadays the most common
geomatics techniques used for 3D
documentation, reconstruction and
interpretation process in the
archaeological field are based on
image data (e.g. photogrammetry) or
range data (e.g. active sensors such
as laser scanners). Both approaches
have their own advantages and
disadvantages and generally the
choice between them is made
according to the budget, project size
and goal, required degree of detail
Laser Scanner and experience of the working team.

5. Image-Based Modelling and Structure from Motion
Nowadays 3D scanners are also becoming a
standard source for input data in many
application areas, but the modern techniques
of Structure from Motion (SfM) and Image-
Based Modelling (IBM) open new perspectives
in the field of archaeological documentation,
providing a simple and accurate way in
recording three-dimensional data.
In computer graphics and computer vision,
Structure from Motion (SfM) and Image-Based
Modelling (IBM) methods rely on a set of two-
dimensional images of a scene to generate a
three-dimensional model.
Compared to laser scanners, the main
advantages of SfM and IBM are that the
sensors are generally cheaper and portable
and that 3D information can be accurately
recovered regardless of the size of the object.

6. Python Photogrammetry Toolbox (PPT)
From a set of images…
The Python Photogrammetry Toolbox
(PPT) is an open source system that
implements a pipeline to perform 3D
reconstruction from a set of pictures.
It takes pictures as input and performs
automatically the 3D reconstruction for
the images for which 3D registration is
possible.
It is done by identifying similar content
between N images and solve 3D
geometry problems.
User input consist of an image collection
and camera parameters. The computed
output is a 3D points cloud.
…to a 3D points cloud

7. Python Photogrammetry Toolbox (PPT) and its Graphical User Interface (PPT-GUI) is
developed by Alessandro Bezzi (ArcTeam – Trento, Italy) and Pier Moulon
(IMAGINE/LIGM, University Paris Est & Mikros Image).
The suite Python Photogrammetry Toolbox (PPT) is composed of python scripts that
automate the different steps of the workflow. The entire process is reduced in two
commands: camera calibration and dense 3D points cloud reconstruction.
PPT-GUI is the
graphical interface to
interact easily with
the photogrammetry
toolbox.
The interface is
designed in two
different parts: a
main window
composed by
numbered panels
which allows the user
to understand the
steps to perform…

8. …and a terminal
window in which the
process is running.

9. RunBundler performs
the camera calibration
step.
Bundler is a structure-
from-motion (SfM)
system written in C
and C++ for
unordered image
collections.
It computes the 3D
camera pose from a
set of images. It
needs only two
parameters: camera
model and sensor
width size.

10. Despite the automation, the user can control the final result choosing two initial
parameters: the image size and the feature detector.
A reduction of the
computation time and
a decreasing density
of the 3D points cloud
depend from the
setting of the first
parameter.
It is a scaling factor
of the image size.

11. Despite the automation, the user can control the final result choosing two initial
parameters: the image size and the feature detector.
The final result
depend from the
setting of the
feature detector:
PPT can work both
with SIFT (patent of
the University of
British Columbia -
freely usable only
for research
purpose) and with
VLFEAT (released
under GPL v.2
license). The second
one is completely
open source, it
ensures a more
accurate result, but
it increases the time
of calculation.

12. RunCMVS takes the
output of structure-
from-motion (SfM)
software as input, and
perform the dense 3D
point cloud
computation.
Data conversion from
Bundler format to
CMVS/PMVS format is
made by using
Bundle2PMVS and
RadialUndistort.

13. Dense computation is
done by PMVS as well
as CMVS, that is an
optional process that
divides the input
scene in many smaller
instance making the
process of dense
reconstruction faster.
The main drawback of PPT is that computation speed depends from user's
computer. There could be some more drawbacks for large scenes or large images
but a compromise between performance and quality can be made by reducing
image size with the scaling factor. Generally, the amount of time needed for the
processing of image sets is in the order of hours.

14. MeshLab
The 3D points cloud processed by PPT is displayed and processed in MeshLab.
MeshLab is an open source system to create a 3D surface (mesh) from a 3D points
cloud (http://meshlab.sourceforge.net/).
MeshLab is developed by Visual Computing Lab of Rome (ISTI-CNR) and is designed
with the following primary objectives:
• ease of use. The tool should be
designed to facilitate users without
high 3D modeling skills
• single mesh processing oriented.
The system should try to stay
focused on mesh processing instead
of mesh editing and mesh design
where a number of other applications
exist a (notably blender, 3D Max,
Maya, and many others)
• efficiency. 3D scanning mesh can
easily be composed by millions of
primitives (points…), so the tool
should be able to manage them

15. From a set of images…
…to a 3D points cloud with
PPT
…to a 3D model with
MeshLab.

16. Points cloud… …mesh…
The point cloud can be cleaned of
outliers, cropped to the area of
interest and then triangulated using
one of the merging filters available in
MeshLab.
The result is a 3D surface. The colour
of the surface (texture) may be
derived from the 3D points cloud,
which is usually coloured.
…3D model with photorealistic texture.

17. Step 1: import the points cloud

18. Step 2: cleaning the points cloud

19. Step 3: surface reconstruction (Poisson or Ball Pivoting approach)

20. Step 4: coloring the mesh using the color of the 3D points cloud

21. The 3D model

22. Archaeological applications: Finds
Archaeological finds can be documented “in situ”, taking pictures moving around
the object, or in laboratory.
In this case 9 pictures were taken with a NIKON Coolpix L110 at 12 MP.

23. Archaeological applications: Finds
Archaeological finds can be documented “in situ”, taking pictures moving around
the object, or in laboratory.
In this case 9 pictures were taken with a NIKON Coolpix L110 at 12 MP.
In PPT the image were elaborated In MeshLab the mesh was created
with the feature detector VLFEAT and with the Poisson surface
with a scaling factor of 0.75. reconstruction filter (octree depth 10,
solver divide 9) and the coloured.

24. 3 cm
The real object

25. 3 cm
The real object
excellent quality for both of details and texture

26. Archaeological applications: Layers
Archaeological field activity is mainly a working process which ends, in most
cases, with the complete destruction of the site. Since most of the interpretation
is performed in a second stage, it is necessary to collect a massive amount of
documentation (images, sketches, notes, measurements).
In the lack of particular expensive equipment (laser scanner, calibrated camera)
or software (photogrammetric applications), field documentation is composed by
pictures, manual drawings, total station measurements and photo-mosaics. While
3D scanning technologies are able to precisely capture the geometry of the
excavation, their cost and operational time discourage an intensive field use.
The alternative is represented by Image-Based Modelling and Structure from
Motion technologies, which are able to obtain a 3D model of an element starting
from a set of images. Using the same instruments of standard archaeological
documentation (digital camera and total station) it is possible to record also the
morphology of the level. The data acquisition is fast and simple and it consists
exclusively in taking pictures of the area of interest. The same rules of the
traditional photography are to be followed: centre the desired object in each
picture, avoid extreme contrast shadow/sun, use a tripod in low-light condition.
There are no limits about the number of images: but it depends on the complexity
of the surface and on the power of the hardware (RAM) which will process the
data.

27. Archaeological applications: Layers - Example 1
In this case 69 pictures were taken with a Sony Cyber-Shot DSC-W90 at 10 MP.
… and many others…

28. Archaeological applications: Layers - Example 1
In this case 69 pictures were taken with a Sony Cyber-Shot DSC-W90 at 10 MP.
… and many others…
In PPT the image were elaborated with In MeshLab the mesh was created
the feature detector VLFEAT and with with the Poisson surface
no scaling factor. reconstruction filter (octree depth
10, solver divide 9).

29. Very good quality of the model and of the details …
Isernia La Pineta, Italy (excavation by prof. C. Peretto, University of Ferrara)

30. … and good texture, very photorealistic.
Isernia La Pineta, Italy (excavation by prof. C. Peretto, University of Ferrara)

31. Archaeological applications: Layers - Example 2
In this case 21 pictures were taken with a Sony Cyber-Shot DSC-W90 at 10 MP.
… and many others…

32. Archaeological applications: Layers - Example 2
In this case 21 pictures were taken with a Sony Cyber-Shot DSC-W90 at 10 MP.
… and many others…
In PPT the image were elaborated In MeshLab the mesh was created with
with the feature detector VLFEAT the Poisson surface reconstruction filter
and with a scaling factor of 0.75. (octree depth 10, solver divide 9)

33. Not very good quality of the texture: the color is too flat
because of poor lighting, but…
Isernia La Pineta, Italy (excavation by prof. C. Peretto, University of Ferrara)

34. … very good quality of the model and of the details.
Isernia La Pineta, Italy (excavation by prof. C. Peretto, University of Ferrara)

35. … very good quality of the model and of the details.
Isernia La Pineta, Italy (excavation by prof. C. Peretto, University of Ferrara)

36. Archaeological applications: Layers - Example 3
In this case 227 pictures were taken with a Apple iPhone 4S at 8 MP.
… and many others…

37. Archaeological applications: Layers - Example 3
In this case 227 pictures were taken with a Apple iPhone 4S at 8 MP.
… and many others…
In PPT the image were
elaborated with the
feature detector VLFEAT
and with a scaling factor
of 0.25.
In MeshLab the mesh was
created with the Poisson
surface reconstruction
filter (octree depth 10,
solver divide 9)

38. The site
Isernia La Pineta, Italy (excavation by prof. C. Peretto, University of Ferrara)

39. The site
Good result of the details
(but centimetric precision,
not millimetric), but not
very good quality of the
texture because of poor
light
Isernia La Pineta, Italy (excavation by prof. C. Peretto, University of Ferrara)

40. Archaeological applications: UAV Aerial photography
A set of photographs taken with a UAV at different flight
altitude (20 m, 35 m and 50 m) are being developed in
collaboration with dr.ssa Paola Piani and Geographike
s.r.l. (www.geographike.it)
The 3D points cloud
The 3D model

41. Conclusions
These experiments demonstrated the possibility to document, in three dimensions
and with a very low budget, archaeological sites and finds.
Advantages:
•fast acquisition and processing
•good scalability, both small and huge model can be acquired
•non-expert users can create his/her 3D model
•cheap
•it needs only the equipment which is normally used in an excavation (digital
camera and total station)
•easily portable hardware components allow archaeologists to work under critical
or extreme conditions (e.g. in high mountain, underwater or inside a cave)
Disadvantages:
•accuracy depends on some factors, in particular the lighting conditions and the
performance of the PC used in the processing
•the texture obtained by the points cloud is very good for objects of small
dimensions, but less for the large models

42. On-line resource
opentechne.wordpress.com
This is our blog where you can find and read tutorials and case
studies about the use of open source applications in archeology
and cultural heritage.
In the download section is possible to download some of our
papers, posters, presentations ...
and this presentation is available for download too.
References
•Gonizzi Barsanti S., Gherdevich D., Degrassi D., 2011. Use Of Low Cost UAV Systems In Archaeological
Research And Disclosure, ISPRS WG V/2 York, UK, Workshop (17 -19 August 2011)
www.academia.edu/1122557/USE_OF_LOW_COST_UAV_SYSTEMS_IN_ARCHEAOLOGICAL_RESEARCH_AND_DISCLOSURE
•Callieri M., Dell'Unto N., Dellepiane M., Scopigno R., Soderberg B., Larsson L. 2011. Documentation and
Interpretation of an Archeological Excavation: an Experience with Dense Stereo Reconstruction Tools.
VAST11: The 12th International Symposium on Virtual Reality, Archaeology and Intelligent Cultural
Heritage: 33-40
http://vcg.isti.cnr.it/Publications/2011/CDDSSL11/Callieri_etAl_Documenting.pdf
•Moulon P., Bezzi A., 2011. Python Photogrammetry Toolbox: una soluzione libera per la documentazione
tridimensionale, Archeofoss 2011. Open Source, Free Software e Open Format nei processi di ricerca
archeologica VI Workshop (Napoli, 9/10 giugno 2011)
http://imagine.enpc.fr/publications/papers/ARCHEOFOSS.pdf (english version)
This work is licensed under the Creative Commons Attribution-Non Commercial–No Derivs 3.0 Unported License.