Things on strings and complex computer algorithms

Abstract

Kite Aerial Photography (KAP) is applied as a tool for documenting excavations at the Tulul adh-Dhahab site in the lower Zarqa valley, northwestern Jordan. The resulting imagery as well as aerial photos from the 1950ies are used to generate 3D models, orthophotos and DSMs by Structure from Motion Photogrammetry (SfM).

Full text

Number 45 September 2012
Contents
Frontispiece: London Olympics 2012 by Otto Braasch 2
Editorial 3
Chairman’s Piece by Oscar Aldred 6
Aerial Archaeology course, Czech Republic, 18-23 June 2012 by Lucie Culikova 9
A picture is worth a thousand words…at least at the Aerial Archaeology
Training School in Merida! by Cristina Charro Lobato 11
Report on summer school of “Potential of satellite images and hyper-/multispectral
recording in archaeology” 30th July – 3rd August 2012 by Jitka Jizerova 14
First Aerial Archaeology Research and Training School in Turkey
20 – 30th July 2012 a.k.a. No Sick Bag For You by Marko Ceković 17
Forthcoming workshop and conferences 19
Four photos taken from a balloon over Berlin in the infancy of aerial
photography by Ole Risbøl and Susanne Kaun 20
Things on strings and complex computer algorithms: Kite Aerial Photography and
Structure from Motion Photogrammetry at the Tulul adh-Dhahab, Jordan
by Jochen Reinhard 37
What a difference a year makes: preliminary DART datasets from Cherry
Copse, Cirencester by R. Fry, D. Stott, D. Boddice and A. R. Beck 42
Pickering’s packages: some thoughts on cursus monuments by Kenneth Brophy 48
Aerial reconnaissance of maritime landscapes in Scotland – some preliminary
observations on context, methodology and results
by Dave Cowley, Jonathan Benjamin and Colin Martin 64
Aerial Archaeology in Jordan 2010-2012 by Bob Bewley, David Kennedy,
Rebecca Banks and Mat Dalton 74
Cropmarks 82
Information for contributors 83
Review article (Historic Wales from the Air) by Oscar Aldred 84
Books of interest? 86
Rebecca Bennett, Kate Welham, Ross A. Hill and Andrew Ford. A Comparison of
Visualization Techniques for Models Created from Airborne Laser Scanned Data
Arne Ramisch, Wiebke Bebermeier, Kai Hartmann, Brigitta Schütt, Nicole Alexanian.
Fractals in topography: Application to geoarchaeological studies in the surroundings
of the necropolis of Dahshur, Egypt
David L. Kennedy. Pioneers above Jordan: revealing a prehistoric landscape
David C. Cowley and Birger B. Stichelbaut. Historic Aerial Photographic
Archives for European Archaeology
AARG: general information, membership, addresses, student bursaries 88
ISSN 1756-753X
A
ARGnews
The newsletter of the Aerial Archaeology Research Group
Photo copyright © Rog Palmer: 17 May 2012

AARGnews 45 (September 2012)
37
Things on strings and complex computer algorithms
Kite Aerial Photography and Structure from Motion Photogrammetry
at the Tulul adh-Dhahab, Jordan
Jochen Reinhard1
What to do when your next excavation project needs proper photographic overview
documentation but the topographic situation of the site does not allow access by either fire
trucks or cherry pickers or similar heavy equipment? Helicopters are of course way too
expensive, because there's no budget to speak of. And to make things even more complicated
your excavation site lies in Jordan, only about 10 kilometers east of the Israeli border –
anything RC controlled will probably lead to serious trouble, an autonomous UAV is nearly
unthinkable. Well, you can always reuse your childhood toys...
In 2008 I was offered the opportunity of participating in an excavation project led by Prof. Dr.
Thomas Pola of the Technical University of Dortmund, Germany. The project focuses on
interdisciplinary research on the Tulul adh-Dhahab: two neighbouring peaks in the lower
Zarqa river valley in north-western Jordan. Inhabited since at least the Bronze Age until Late
Antiquity they comprise a wealth of archaeological traces, numbers of which are still
observable on the surface (Gordon, Villiers 1983; see also http://en.wikipedia.org/wiki/Tulul_adh-
Dhahab, where some of the aerial imagery shot during the 2008 campaign is also included).
The yearly field campaigns, since 2011 directed by Hannelore Kröger, take place on the
topmost plateau of the western tell: the Dhahab el-Gharbiyeh site.2 To overcome the
photographic problems described above, Kite Aerial Photography (KAP) seemed to be the
best solution because of its simplicity, inconspicuousness and transportability. Basically all
you need is a kite, a camera and a clever device called a rig which holds the camera, keeps it
steady and points it in the desired direction.
Lifting cameras with kites is not a new idea, first successful images were obtained by the
French Arthur Batut in 1888 (Batut 1890; Autha et al. 1988). Although Batut already
considered an application of this new technique to archaeology, the first archaeological use of
Kite Aerial Photography was initiated by Sir Henry Wellcome and executed by A. G. Barrett
at the 1912/13 campaign at Jebel Moya in Sudan (Addison 1949, e.g. 6) – hey, we are nearing
the centenary jubilee! Since its invention, researchers have made regular use of Kite Aerial
Photography in various scientific disciplines including archaeology (Żurawski 1993; Tielkes
2003; Verhoeven 2009, 237 f.; Aber et al. 2010, 103-109; Bibi et al. 2012). It has never been
a common standard method, but lately the emergence of digital photography has led to a new
rise in scientific kiting.
Having decided to give KAP a try I set out to get the necessary equipment: a 9 megapixel
digital compact camera, a FUJIFILM FinePix E900, was already in my possession – and I
was mercilessly willing to risk smashing it on the rocky slopes of the Jordanian hills. (My
wife maybe was not so willing, but in the end the camera lived to tell the tale!) As the kites of
my childhood seemed a bit on the small side for lifting a camera I bought an Invento HQ
Flowform 4.0. The equipment was completed with 100 meters of kite line, a winder and a
1 jochenreinhard@web.de
2 JADIS: # 2117003, MEGA: # 2709; see http://www.megajordan.org/.

AARGnews 45 (September 2012)
38
Brooxes Servo-BEAK rig kit (http://www.brooxes.com; http://www.kapshop.com). This
device hangs from the kite line by a gimbal contraption made of string called a picavet and
triggers the camera via a small servo actuated lever pushing on the shutter release button
about every 10 seconds (Figure 1). Adapting to the needs of the Tulul adh-Dhahab excavation
the equipment was intentionally kept as simple as possible. Of course the last bits of
equipment arrived only days before departure so prior testing had to be cancelled, everything
had to be “tested on the job”.
And the testing went well: Even the very first KAP flight yielded good imagery! In only five
flights of about 10 to 25 minutes (plus a few unsuccessful low wind attempts) we3 succeeded
in getting vertical shots of all our trenches during and at the end of the dig and an additional
number of low oblique images documenting the work on the ongoing excavation (Figures 2-
3). The application of the method is simple: Start the kite, let out line until it flies safely, put
3 Many hearty thanks to Christiane Kühne, Christian Günther, Gerke Sedat and to my wife, Miryam, for
helping to navigate the rig above the wanted motif, sorting out kilometers of tangled line and rescuing rig and
camera more than once!
Figure1‐PreparingtolaunchthecamerarigonthetopofTalladhDhahabel‐Gharbiyeh(left,photobyHanneloreKröger).
The2008KAPsysteminflightwiththecamerarigonthelinebelowthe4m2HQflowformkite(middle,photobyTatiana
Schär).CloseupoftheCanonG11camerarigasusedin2011(right).
Figure2‐Obliqueviewofthe2011excavationworksat
theuppermostterraceofTalladhDhahabel‐Gharbiyeh.
InthebackgroundthedeepvalleyoftheZarqariver.
Viewtoeast‐south‐east.
Figure3‐Verticalviewofthe2011excavationtrenches
attheuppermostterraceofTalladhDhahabel‐
Gharbiyeh(seefigure2).Therectangulartrenches
measureapprox.3.7x3.7m,theirregularholesare
mostlyrobbertrenches.

AARGnews 45 (September 2012)
39
on the camera rig, let out more line until the rig reaches the desired flying height, adjust rig
position by walking around (a second person functioning as a navigator is very helpful here)
and let the intervalometer do the work while enjoying flying the kite.
The surprisingly good and easily achieved results encouraged me to pursue the project. When
I attended the dig the next time in 2011, my equipment had been thoroughly upgraded: The
old camera had been replaced by a 10 megapixel Canon PowerShot G11 with image
stabilisation running a CHDK intervalometer script (http://chdk.wikia.com/wiki/CHDK), the
rig had been fitted with legs and little ball-bearing blocks to provide less friction for the
picavet lines and the old flowform kite was substituted by a Gomberg Dazzle Delta kite with a
higher flying angle and far less pull. Together with a better routine these improvements
resulted in a considerably lower percentage of unusable pictures and a distinctively better
image quality. Additionally a Holux M-241 GPS logger was flown on the kite line directly
above the rig to get a rough measurement of the camera location during flight.
The wind situation at the Tulul adh-Dhahab is generally very stable in late summer: Only
occasionally a light eastern morning breeze is strong enough for getting the 500 or 650 grams
of KAP rig airborne and taking a few photos, but normally it dies down shortly after sunrise.
A bigger kite or less rig weight could help here. During the day the wind usually is quite
calm, so taking pictures with a high sun is hardly possible. But around late afternoon a
moderate to fresh – and sometimes even strong – western breeze very reliably sets in, blowing
up the Zarqa river valley. This wind is perfectly suited for getting low altitude aerial shots of
virtually all archaeological remains on the hill. At this time of the day, shadows already grow
longer, but the less harsh evening light is much better suited to photography than the midday
glare.
The camera height usually varied between 20 and 60 meters, allowing for a field of view
between 19 x 25 and 56 x 75 meters with ground sample distances of 6.8 to 20.5 mm
respectively – but of course a kite’s flying height is only a matter of line length. The cameras
were set to the highest JPEG quality setting and minimal focal length as more detail is much
easier achieved by flying lower than by zooming in. To counteract motion blur all pictures
were shot in shutter priority mode with the fastest shutter speed possible while keeping the
ISO value low. The images provide an
excellent addition to the ground based
excavation photography as they supplement
unique vertical and oblique perspectives and
give a great overview otherwise only
obtainable with much higher costs and hassle.
Even after the advent of autonomous camera
carrying drones Kite Aerial Photography
proved to be a valuable tool in the aerial
archaeologist's tool kit – and it’s fun, too!
Another bonus is that KAP is not only
valuable for documentation purposes: While
shooting verticals of the ramparts at the tell’s
northwestern flank the kite suddenly pulled
sidewards in a gust – and inspecting the
Figure4‐AccidentalKAPsurvey.Apreviouslyunknown
structurecouldbedetectedaftera2008KAPflightinthe
upperrightoftheimage.Therampart'stowercanhardlybe
seenatgroundlevelbelowthefootpathleadingtothehilltop.

AARGnews 45 (September 2012)
40
photos back in the camp I discovered to my surprise that the image taken just in this moment
showed a rectangular structure previously unknown to us (Figure 4). The structure, the ruin of
one of the rampart’s towers, is almost impossible to see at ground level between the rubble of
the tumbled down walls and several robber trenches, even if it's existence and position is
known. So kites are also valuable tools for survey work – on a different scale to aircraft of
course.
The valuable images of the KAP flights above the Tulul adh-Dhahab were improved even
further when I by chance stumbled upon the Agisoft PhotoScan software
(http://www.agisoft.ru/; Verhoeven 2011; Verhoeven et al. 2012a; 2012b): Suddenly the
superfluous pictures recorded by the intervalometer trigger weren't superfluous at all –
suddenly they provided necessary overlap for the Structure from Motion algorithms
PhotoScan uses to calculate its 3D point clouds! And the unordered, unsystematic vertical kite
imagery seemed to be ideal for this kind of photogrammetry as the lightly swaying kite
provides imagery with different angles only slightly deviating from nadir but providing lots of
overlap (Figure 5 left and middle). Virtually every single picture of a successful KAP flight
can be incorporated in the final 3D model. Georeferenced to the excavation grid this model
provides not only a precise DSM of the trenches and their surroundings but the professional
version of PhotoScan also allows the export of true orthophotos, too!
However the use of Agisoft PhotoScan at the Tulul adh-Dhahab is not limited to kite aerial
imagery: In 1953 the British Hunting Aerosurveys Ltd. flew an aerial photography mission
over western Jordan (e.g. Kennedy, Bewley 2009, 70). The resulting imagery, some 4.000
vertical images in 1:25.000 scale, is held in the ‘Royal Jordanian Geographic Centre’ in
Amman. The ‘Aerial Photographic Archive for Archaeology in the Middle East’ of the
University of Western Australia (http://www.apaame.org/) is in possession of a set of copies
and thankfully provided me with the necessary high resolution scans. PhotoScan had no
problems calculating the camera positions and produced a nice DSM of the Tulul adh-Dhahab
area. Astonishingly the flight path of the Huntings aircraft nearly 60 years ago could be
reconstructed with ease (figure 5 right). What makes these DSMs valuable for archaeological
work is that by using current SfM software we are able to reconstruct a topography now
severely changed by erosion and bulldozers.
And while you have read all this, the author hopefully is again flying one of his kites over the
Jordanian desert – of course with a camera dangling from its line...
Figure5‐PhotoScanscreenshotof82alignedkiteaerialphotographsfromasingleKAPflightoverterraceI,campaign2011,
seeFigures2and3.Theblueplanesrepresentthecamerapositionforeachofthephotosusedtocalculatethe3Dmodel.
Notetheunorderedcharacteroftheimageryandthedifferentcameraheightsresultinginavariedratioofdetailand
overview.Viewtothewest(left).Exportedorthophotoofthemodelontheleft,nottoscale.Northisup(middle).PhotoScan
screenshotofalignedaerialphotographsofthe1953HuntingsAerialSurvey,northisup,theareashownisapprox.14km
wide.Theaircraft'sflightpathcanbereconstructedaccordingtothenumberingofthephotographs(right).

AARGnews 45 (September 2012)
41
References
Aber J. S., Marzolff I., Ries J. B. 2010. Small-Format Aerial Photography. Principles, Techniques and
Geoscience Applications, Amsterdam 2010.
Addison F. 1949. Jebel Moya. The Wellcome Excavations in the Sudan I+II, London 1949.
Autha D., Negre S., de Beauffort G., Fosset R. 1988. Labruguière. Berceau de l'aérophotographie par cerf-
volant – birthplace of kite aerophotography. Arthur Batut 1846-1918, La Primaube-Aveyron 1988.
Batut A. 1890. La Photographie Aérienne par Cerf-volant. Bibliothèque photographique, Paris 1890.
Bibi F., Kraatz B., Craig N., Beech M., Schuster M., Hill A. 2012. Early evidence for complex social structure in
Proboscidea from a late Miocene trackway site in the United Arab Emirates. Biology Letters, published
online before print on February 22, 2012, doi: 10.1098/rsbl.2011.1185 [see also
http://mleisa1.wordpress.com/ (accessed July 8, 2012)].
Gordon R. L., Villiers L. E. 1983. Telul edh Dhahab and its Environs Surveys of 1980 and 1982. A preliminary
report. Annual of the Department of Antiquities of Jordan 27, 1983, 275-289.
Kennedy D., Bewley R. 2009. Aerial archaeology in Jordan. Antiquity 83/319, 2009, 69-81.
Tielkes E. 2003. L'œil du cerf-volant. Evaluation et suivi des états de surface par photographie aérienne sous
cerf-volant, Weikersheim 2003.
Verhoeven G. 2009. Providing an Archaeological Bird’s-eye View – an Overall Picture of Groundbased Means
to Execute Low-altitude Aerial Photography (LAAP) in Archaeology. Archaeological Prospection 16,
2009, 233-249.
Verhoeven G. 2011. Software review. Taking Computer Vision Aloft – Archaeological Three-dimensional
Reconstructions from Aerial Photographs with PhotoScan. Archaeological Prospection 18/1, 2011, 67-
73.
Verhoeven G., Doneus M., Briese Ch. 2012a. Computer vision techniques: towards automated orthophoto
production. AARGnews 44, 2012, 8-11.
Verhoeven G., Doneus M., Briese Ch., Vermeulen F. 2012b. Mapping by matching - A computer vision-based
approach to fast and accurate georeferencing of archaeological aerial photographs. Journal of
Archaeological Science 39, 2012, 2060-2070.
Żurawski B. 1993. Low altitude aerial photography in archaeological fieldwork: the case of Nubia.
Archaeologia Polona 31, 1993, 243-256.