Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Methodological contest between high-end hardware and low-cost
equipment for archaeological documentation
keywordsstructure from motion, image-based modeling approach, 3D geo-spatial field technology, archaeological data acquisition, 3D laser scan, FOSS, GRASS GIS, high resolution digital documentation technique
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction -> Research question, State-of-the-art of digital documentation methods
Aim -> Comparison of two methods
Case study -> Ostia Antica
Results -> And the winner is …!
Outlook -> Future perspectives
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
Research question
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
State-of-the-art of digital documentation methods in Prehistoric Archaeology
First application of terrestrial laserscanners in Prehistoric Archaeology by VIAS Wien
Quelle: VIAS Wien
©VIAS Vienna
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
State-of-the-art of digital documentation methods in Prehistoric Archaeology
First application of terrestrial laserscanners in Prehistoric Archaeology by VIAS Wien
(© ArcTron)
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
Advantages & Disadvantages
- High quality and precise documentation- immediate control- Hardware: quick and easy to use- Software: can be operated intuitively
after an introduction- Fast and easy referencing and geo-
referencing possible- Fast data processing = data preparation
- Time consuming because of mostly more than one scanworld is needed
- Every scanworld needs a new registration by scanning all targets = time consuming
- Hardware: havy equipment- Difficult custum regulations in
certain countries- Very expensive equipment- Software: Very expensive
proprietary for documentation and first data preparation
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
State-of-the-art of digital documentation methods in archaeology
Computer Vision / Structure-from-motion
- originally developed for Tourism / Social networking purposes
- Online-Services like Photosynth or Photofly for all all kinds of use:
private, construction, forensics, design…
-> ca. 150 industries using it (NY Times)
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Computer Vision / Structure-from-motion
- Adaption for archaeology as a low-cost alternative to laser scanning
- different packages available:
- Proprietrary: Photoplan, Arctron aspect3D
- Open source: ArchOS, BundlerTools, VisualSFM
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
Advantages & Disadvantages
Advantages:
- little gear
- low-cost
- time-saving (on-field)
- more weather resistent
- flexible
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Disadvantages
- spatial cover
- post-processing (Workflow)
- no immediate control
- required compute power (but no cloud-computing necessary)!
- georeferencing
Advantages & Disadvantages
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
Comparison of two methods
Laserscanning Structure-from-MotionHardwarePrice High LowTransport Heavy and difficult Just one photo camera ist neededAdditional Equipment + Tripod, + Notebook, + certain
targetsNo
Power supply Needed NeededAcquisition time 6x as long as SfMWorkflow stepsdistance, spatial cover The longer the distance, the
lower the resolution .
field of application Architectur, Prehistoric Excavation.
Weather sensibility Not applicable by rain and temperature higher 40°C and
lower – 40°C.
Handling management Difficult to handle
SoftwarePrice highUsability can be operated
intuitively after an introduction
MaintenanceProcessing timePost-processing
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
OfficeFreyensteinOstia Antica
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
Ostia Antica
1. 489.381, 702.880, 2.747
2. 490.217, 701.778, 2.813
3. 488.992, 702.380, 2.844
4. 490.498, 702.327, 2.781
1. 489.376, 702.918, 2.725
2. 490.206, 701.811, 2.813
3. 488.969, 702.372, 2.857
4. 490.475, 702.319, 2.767
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
164 743 vertices
Laserscan (resolution 1 mm, 1 Scanworld)
Structure-from-motion
238 247 vertices
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
Ostia Antica
• Taking pictures: easy
• Taking usable pictures, some things to think of:
• Good camera objective helps
• large memory cards necessary
• Don´t forget targets for measurements and
color balance
• more cameras would even speed up
• assure the overlap!
• Laptop on site for instant check of photos and
running of small models
Data acquisition of structure from motion
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
Ostia Antica
Open-source: bundle of software needed:
• Image viewer/Graphics software -> sort / rename images
• Linux: BundlerTools -> create pointcloud
• Meshlab, CloudCompare -> clean pointcloud, pick points
NEW: GRASS GIS:
.ply-Import/Export module + georeferncing (rectify)
• Paraview, (AutoCAD), Meshlab, CloudCompare, Blender -> analyzing
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
3D Georeferencing application flow:
1) Import of the point cloud in different file formats: e.g. PLY, LASPLY Vertex-attributes included
v.ply.in
2) Georeferencing of vectors in GRASS GISv.transform
3) Export of the point clould in different file formats: e.g. PLY, VTKv.ply.out
Maximum size of the pointcloud: 2 billion points
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
3D Georeferencing
Affine and higher polynomial transformation based on control points
The accuracy of the transformation is depended on the accuracy of the control points
Affine transformation:For each dimension separately: move, scale, rotate
Helmert-Transformation (coming soon!):For each dimension separately: move, rotateFor all dimensions equally: scale
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
And the winner is …!
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Introduction
Aim
Case study
Results
Outlook
Future perspectives
Computer Vision / SfM (Open-source)
• Toolchain from digital images
to georeferenced 3D-model
available
Things to improve:
• Usability of software packages
- automatic target detection
• Database connection
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Undine Lieberwirth, Bernhard Fritsch, Markus Neteler, Markus Metz
Co-operation between the
Excellence Cluster 264 - TOPOI
Berlin/Germany
and
Fondazione Edmund Mach –
Research and Innovation Centre
Trento/Italiy
THANK YOU FOR YOUR ATTENTION!