You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.


The use of close-range photogrammetry in
zooarchaeology: Creating accurate 3D models of
wolf crania to study dog...

Article in Journal of Archaeological Science: Reports · October 2016
DOI: 10.1016/j.jasrep.2016.06.028


5 359

9 authors, including:

Allowen Evin Greger Larson
CNRS University of Oxford


Thomas Cucchi Keith Dobney
French National Centre for Scientific Research University of Liverpool


Some of the authors of this publication are also working on these related projects:

Palaeomicrobiome View project

The pseudosuchians' bone ornamentation: Morphology, Function(s), Evolution. View project

All content following this page was uploaded by Carly Ameen on 19 July 2016.

The user has requested enhancement of the downloaded file.

This is an open access article under the CC BY-NC-ND license (http://creativecommons. 2014. Oxford OX1 3QY.. Porter et al. 2013. Richard Allen d. requiring only a conventional camera and grammetry have recently experienced a resurgence of interest. photogrammetry has been used In parallel with the rapid development and use of 3D model ac- quisition. both cheap and portable. 34095 Montpellier Cedex 05. Elphinstone Road. . particularly in archaeological research.are becoming common tools for construct- Received 17 February 2016 ing three-dimensional (3D) models of artifacts. geometric morphometric approaches are also increasingly ⁎ Corresponding author at: Institut des Sciences de l' 2013. 2014. 2013. Evin). it remains relatively expensive and not always readily the distinct advantage over other 3D model building methods . Haukaas and Hodgetts. Evin et al. 2014. 2016).95%).elsevier. cade. Introduction intensively in a geographic context . Archéozoologie.1016/j. The latter.being portable. photogrammetry is now becoming a common tool in archaeo- Whilst surface scanning technology has revolutionised the acquisition logical research. geometry.. Newton et al.04% and 1. texture and most Keywords: importantly. 75005 Paris. Verhoeven et al.0/). Keith Dobney b. ning in the mid-nineteenth century. 2009. University of Aberdeen. a geometric comparison computed the deviation map between the pairs of 3D models. Ros et al. This is an open access article under the CC BY-NC-ND license (http://creativecommons.0/). IRD. Greger Larson d. the quantitative comparison of the models revealed an average distance between the two surfaces of 0. affording advantages that make it a highly useful tool for zooarchaeological research. the more traditional techniques of photo. Archéobotanique Sociétés. we explore whether 3D models of wolf crania. face digitisation of objects using close-range techniques over the last 2016. 2011.2016. obtained through a Available online xxxx photogrammetric approach. E-mail address: allowen.evin@univ-montp2.⁎. Published by Elsevier Ltd.. EPHE. Begin. Whether models obtained Received in revised form 29 May 2016 through photogrammetry can be used for zooarchaeological studies requires a systematic examination. In the Accepted 14 June 2016 context of research into dog domestication. colouration and geometry. Bouby et Montpellier.doi. L69 7WZ.b. 2012. Université de being applied to address bioarchaeological questions (e.088 mm with an average standard deviation of 0.g.c. Classics and Egyptology.. offers a low cost. Pratiques et Environnement. UK d Palaeogenomics & Bio-Archaeology Research Network. we compared the topology of 3D models obtained with a high-resolution surface scan- Skull ner (used as reference geometry) with models reconstructed from the same five wolf crania using photogram- Canis lupus metry. Sapirstein. In addition. Counts et al. These close-range techniques of photogrammetry have of 3D models. Carly Ameen b. CNRS. Published by Elsevier Ltd. Grosman et al..jasrep. Place Eugène Bataillon. Thibaud Souter b. Geometric morphometrics To answer this question. Ardern Hulme-Beaman b.. Dyson Perrins Building.. 12-14 Abercromby Square.. Liverpool.. CNRS. survey. Mcpherron et with only 6. © 2016 The Authors. France. University of Liverpool. Université de Montpellier. 55 rue Buffon. easily portable and simple to perform alternative to tra- ditional surface scanning. 2007. 2016). As a consequence. al. Journal of Archaeological Science: Reports 9 (2016) 87–93 Contents lists available at ScienceDirect Journal of Archaeological Science: Reports journal homepage: (A. Surface scanning which was then followed by a 3D landmark based geometric morphometric approach using corresponding 3D models analyses. The pairs of models were then compared using both a visual.53 mm.. Place Eugène Bataillon. The geometric morphometric analyses revealed the same degree of measurement error for the two series of scans (2.c. 1. Driven by technical developments for the sur- (e.g. UMR 7209.including photogrammetry . 15 years. UK e CNRS-Muséum National d'Histoire Naturelle. accurately describe the original cranium in term of colouration. therefore.31% of the morphometric variation being due to the acquisition technique. Aberdeen. Pagnoux et al.g. minimal accompanying equipment and setup. Pietro Viacava e. The use of close-range photogrammetry in zooarchaeology: Creating accurate 3D models of wolf crania to study dog domestication Allowen Evin a. 2014. France a r t i c l e i n f o a b s t r a c t Article history: Close-range photographic techniques . Seetah http://dx. 2008. Cucchi et al. Mary's Building. St.c. spatial planning and geomorphological anal- Image-based 3D modeling has grown significantly over the last de. 2013. Pavlidis et al.028 2352-409X/© 2016 The Authors. offering new possibilities for recording archaeological artifacts Yamafune et al. IRD. South Parks Road. 2016. qualitative and two quantitative Wolf approaches. Photogrammetry. Photogrammetry Our results demonstrate that photogrammetry can produce 3D models with visually satisfying levels of morpho- logical detail in terms of texture. yses (e. France b Department of Archaeology. UK c Department of particular for applications relat- ed to cartography.06. EPHE.c a Institut des Sciences de l'Evolution. 34095 Montpellier Cedex 05.b. Núñez et al.. 2016. Thomas Cucchi e.

15° and 40°. which is both a gle-lens reflex (DSLR) Canon EOS 30D camera. com) coupled to its automated turntable. Wayne. 2014. The 2 × 12 set of subsequent morphometric analyses. a 2. With each cranium placed successively tural history of humans. 1991). distinguishing between wild and domestic forms. Wu. see Fig. but these measurements and MNHN-ZM-MO-1997-454.firstly by computing the deviation map be. including concavities.2. In that study. In 2012. and often include as well as unambiguous feature points for image orientation and (essen- a reduction in size. 2009). 2011). This pattern provided a visual aid for 10-degree increments.. StereoScan structured light scanner (http://www. the photographs were shot at regular intervals mesticate was the dog. Tchernov and Kolska Horwitz. 2010).. Gunz and Mitteroecker. 2012. each (Friess. In this set-up.breuckmann.. 2005. Studies combining both photogrammetry and geometric mor. Weinberg et al. 2. Souter et al. Clustering Views for Multi-view Stereo and models were compared using both a visual qualitative approach and Patch Based Multi-view Stereo Software. which entered the human sphere long of approximately 10°. Sets of 36 pictures were acquired from three attempting to separate e. The anatomical complexity of the cranium is challenging for surface Each dorsal and ventral sets of 108 images were imported into reconstruction methods. 1999. 3D views was semi-automatically aligned along the scanning process. Photogrammetry reconstructs 3D models from a set of photographs ferences between individuals than between the technologies used taken from various angles.PLY file. Morey. Katz and Friess. These images must cover the entire surface to construct the models. 2013. The crania were set on a mestication.e.. photographs were centred primarily on the object remains controversial. offering a diagonal scope of 250 mm can then be triangulated . 3D models for use in geometric in human crania by comparing 3D models obtained through a photo. holes or intersecting. This freeware diverse scale and nature. formed using its dedicated software Optocat (http://www.forming a 3D mesh that can be rendered and an average spatial precision of 18 μm. Scanner and photogrammetric reconstructions and texture onto these models can aid accurate identification of the desired landmarks. Evin et al. 2014).e. 1B)..g. points between pairs of photographs). Lowe. sional topology of the external. 2D and 3D sliding in the collections of the Muséum National d'Histoire Naturelle (Paris.with much of it within the (Boudadi-Maligne and Escarguel. on a small central desk.PLY files. two dense 3D point clouds (one for the ventral tween the pairs of 3D models and secondly using a landmark based geo. we compared 3D models of five modern wolf crania ob. sparse 3D points generation metric analyses. Katz and Friess (i. and diffuse light source that reduced the presence of glare or directional.e. Therefore. It is generally accepted that the first do. These dense point clouds required substantial manual cleaning before . The scanner-based 3D models were generated using a Breuckmann graphs taken from multiple angles to reconstruct the three-dimen.breuckmann. MNHN-ZM-MO-1997-453 marks can be obtained from 3D models. The data acquisition was per- as a polygonal model. noise.e. 1A) stressed the need for analysing not only the size of the cranium. whilst also including a fair amount of the surroundings in the ‘domestic’ from either contextual or morphological criteria frame (approximately 2/3rds of the image) . metric morphometric approach and corresponding analyses. (Boudadi-Maligne and Escarguel. 2014. semi-landmarks (Bookstein. Photogrammetric-based 3D models geometric morphometric analysis revealed larger cranial shape dif.. with the most ancient canids identified as of interest. in order to assess in each image) through SIFT algorithms (i. a complete phometrics are rare . Wu et al. planar cardboard sheet displaying a calibrated referential pattern morphological changes are traditionally used as criteria for (Fig. In most cases of animal do. al. Scale Invariant Feature whether photogrammetry can precisely reconstruct high-resolution. as it possesses significant topological detail of a VisualSFM for photogrammetric processing (Wu. 2010). 2011. MNHN-ZM- Gunz et al. These rigid. To ensure proper convergence of the ori- and location of dog domestication from their wolf ancestors entation algorithms. homologous locations – i. Wu. for cranium lying on the lateral side) on the turntable. 1981. a total of 216 (36 × 3 × 2) pictures were used to reconstruct each model. For each specimen. 2007). and one for the dorsal view of the cranium) were saved as . Furukawa and Ponce. tially) fixed dimension geometry to scale the 3D models for later geo- 1992. rendering both colour 2. Material and methods employed to measure and analyse the size and shape of objects using coordinates taken from specific. with surface area measure- ments slightly larger for photogrammetry models. order to assure it did not present any major geometric inconsistencies grammetric protocol and a high precision scanner. 2015. Davis. MNHN-ZM-MO-1997-452. Transforms. requires further examination. 1986).e. 2016) with 12 stops per-pass. In addition.1. Scanner-based 3D models Photogrammetry relies on partially overlapping digital photo. the scanner was mounted with the interest.1. / Journal of Archaeological Science: Reports 9 (2016) 87–93 et al. especially of the skull (e. (Sholts et al. abnormal or non-manifold faces). thereby confirming the appropriateness of the object with significant overlap between pairs of photographs.. 2013. 1A). Geometric morphometrics are 2. Its The images were acquired in a large room offering a very abundant generalization to other mammalian taxa and to other photogram. However. Terral et al. (2014) found a low degree of deviation.1. 2010). EF 24–105 mm f/4 L IS USM lens.1.g.mostly limited to the fields of anthropology specimen scan was obtained in two successive full rotations. MO-1996-2499. Katz and Friess assessed the suitability of photogrammetry before being fused into a single mesh and exported as a . two quantitative methods . are only valid if the said model's accurately retains the geometry of the physical object studied. 1997. the accuracy cally (with the rostrum pointing upward) and horizontally (with the with which photogrammetry preserves the geometry of objects. The timing era around in a circle (Fig. metric protocols now remains to be tested. several studies have different vertical angles (approximately 0°.geomagic. Therefore. depth-of-field and an aperture stop set at f/16. Drake et later scaled and digitally re-assembled into a single object. mounted with a Canon complex and dynamic field pertaining to the more recent bio-cul. when metrical comparison. Coordinates of these land. for both the dorsal and the ventral sides of each cranium. tie-points matching (matching and geometrically accurate. The five specimens of wolf crania included in this study are housed landmarks – or along curves and surfaces – i.. France) – specimens numbered MNHN-ZM-AC-1997-12. as the operator moved the tripod-mounted cam- before the domestication of other plants and animals. which were but also its shape. The pairs of PMVS algorithms (i. In addition. visible surface of an object of Considering the size of the objects.g.88 A. wolves and dogs. The reconstructed 3D coordinates form a ‘point cloud’ that medium-range of optical lenses. The as a tool for capturing and quantifying the morphology of the mesh was then cleaned using Geomagic (www. harsh cast-shadows on the object. of combining photogrammetry with geometric morphometrics. 2011) and finally to reconstruct tained through i) photogrammetry and ii) high-end structured light high-resolution 3D point-clouds by dense correlation through CMVS- surface scanning used as a reference for ‘true’ geometry. Hassett and Lewis-Bale. a number of morphological changes occurred. openings and distinctive allowed us to perform keypoint detection (unambiguous feature-points biological landmarks of various kinds. The crania were successively positioned verti- and medicine (e. We used an 8 mega-pixel digital sin- Our research focuses on animal domestication.

geometric morphometric approach. and the resulting meshes were scaled to the actual and a ‘test geometry’ (here the photogrammetric model) and retrieved dimension of the object using Geomagic. In order to assess the A first comparison of the pairs of models includes: 1) a visual obser. were de. 2003). Rohlf and Slice. B: fixed dimensions reference pattern used to scale the models and enhance the performance of key-points detection/matching and camera calibration algorithms. 1999). 2). the five replicates per model were averaged and a Procrustes their 3D topology. five times per specimen. error linked with the landmark digitisation process. Procrustes Anovas vation of the models. least-square optimisation best-fit alignment) in order to compare Then. 1. with individual as the needed to acquire the photographs (see SI-text and Supplementary EvoMorph) and superimposed using a Generalized Procrustes Analysis Fig. again with individual as the factor. signed and implemented for image acquisition.B. whilst drastically reducing the amount of time ‘Landmark Editor’ (http://graphics. Differences are expressed using the colour scale on the left. 1995. were then aligned. and 2) the computation of a mesh-to-mesh were performed for data acquired on the Breuckmann and deviation map. 3D models comparisons (PCA). This landmark approach involved N. A.ucdavis. (Goodall. The pairs of models were spatially aligned (using a photogrammetric models separately.) . We computed the local distance for each 3D point be. providing models of Landmark coordinates were recorded. A: schematic representation of the set-up and camera positions used for the acquisition of the photographs. recording a set of 28 3D-landmarks on the 10 models using a protocol requiring much less room and fewer operator interventions. 2. A second analysis compared the geometric accuracy of the two before being fused into a single 3D model that then underwent a final acquisition protocols with a three dimensional landmark based cleaning procedure to warrant its geometric consistency. Procrustes Fig. Models obtained with photogrammetry (top) and the Breuckmann structured light scanner (bottom) with the cloud-mesh distances visualisation (middle). (For interpretation of the references to colour in this figure legend. the reader is referred to the web version of this article. 1). and the shape differences associated with the PCA axes visualised using multivariate regressions (Monteiro. The var- iation in shape was synthesized using a Principal Component Analysis 2. Anova performed. new semi-automated and automated protocols.idav. using comparable quality. 1990) and the right and left sides of the crania were symmetrized (Kolamunnage and Kent. adapted from Drake and Klingenberg (2008) (Supplementary Fig. After this study. The two cranium halves the average distance and the standard deviation values. using a least-square optimisation best-fit alignment. Evin et al. being triangulated. in order to quan- tween a ‘reference geometry’ (here the structured light scanner model) tify the part of the variation linked to the scanning technique. / Journal of Archaeological Science: Reports 9 (2016) 87–93 89 Fig.

differences in scanning technologies were vis. In the latter case. 2). In contrast. and 1. The capture and rendering Our data revealed that photogrammetry can be an accurate tool to of topological detail was good and appeared similar in all pairs of 3D reconstruct 3D models of wolf crania in terms of both geometry and ap- models (Fig. therefore. The only areas that differ try offers a very promising tool for geometric morphometric studies of more markedly between the pairs of models (in red or dark blue) lie ei. Results technique. Along these two axes . portability and simplicity. 4. 3).726).61 mm) (Table 1). Photogrammetry also has the advantage of being inherently porta- tween the two surfaces of 0. Both the scannographic and photogrammetric models succeeded in producing 3D models that displayed visually satisfying levels of mor.generally below 0. The landmark digitisation error for the Core Team. low-fidelity colour mapping that software including e. in remote areas or in places offering a limited or unreliable The first two axes of the PCA computed on the shape data represent. This benefit significantly reduces the time axes (in grey) showing proportionally broader brain case and zygomatic devoted to digitisation and thus maximizing the number of models arches (Fig. because of their weight. A significant advantage when targeted specimens cannot ed 75. with most of the differences being Given its lack of expense. in our study the ible for at least three of the specimens (MNHN-ZM-MO-1997-453. ren. and carries its own bat- with an average standard deviation of 0. a feature currently not offered part of the cranium. power supply. The or other institutional collections. 3.95% for the Photogrammetric models and 6. the measurements strongly with the two techniques (using individual and acquisition technique differ between the two scanning technologies (all p b 0. with the specimens in the negative side of the by most surface scanners. pearance.5 mm (Fig. In ad- high amount of radiometric fidelity and consistency. acquisition of all the pictures required for the 3D reconstructions MNHN-ZM-MO-1997-454 and especially MNHN-ZM-MO-1997-452). Photogrammetry is also relatively the colour texture by equalising and averaging pictures taken at each inexpensive .g. importance adminis- 47.g. quantitative comparison of the models revealed an average distance be. provide an equally good alternative to the Breuckmann models presented higher resolution of fine details. as part of a researcher's equipment when visiting remote field locations ative morphometric (including geometric morphometrics) studies. Here. As an example.. methods greatly overlapped with nearly the same range of variation. Procrustes Anovas revealed strong differ- techniques for each of the specimens independently. models are freely available online. tutorials explaining the entire process from photographs to 3D The mesh-to-mesh distance comparisons reveal a high level of sim. Another major advantage of photogrammetry is its potential for and with the measurements of each specimen showing no overlap performing data acquisition (photographs) as a quick and separate (Fig. 4). that can be collected during sessions. Discussion phological detail in terms of texture. Fig. the field. and a two-way ences between the specimens (p = 0. max: tery power supply – all very important assets when working either in 0.47 mm.53 mm (min: 0. blends and softens minor local details like bone sutures and other Photogrammetric 3D models are also relatively easy to obtain. compared with for which measurements corresponding to the photogrammetry the forty-five minutes required on average by the structured light sur- model do not overlap with measurements acquired with the face scanner we used. 2013) and Rmorph (Baylac.90 A. tions for the conservation of cultural heritage and are routinely included sible to surface reconstruction methods and are not included in compar. individuals differ mainly in the shape of the posterior step from the 3D data reconstruction. colouration and geometry (3D models are provided as supplementary files). except when viewing the tooth row. Such highly concave lighting equipment are readily available and commonly used in institu- zones like these are admittedly considered less visible and thus inacces. High-end digital cameras and simple ther within the nasal cavity or occipital foramen. 2012) packages for ‘R’ (R (Supplementary Fig. For each specimen. photogramme- very small . dition. 2012.31% of the variation is explained by the scanning 3. ferences are homogeneous among specimens (p = 0.04 mm. 2014). 4). Breuckmann models is 2.16% of the variance explained . However. The surface scanner produces tured light or laser scanners. 2). Evin et al. as the camera is usually small and durable. since marks. Meshlab (Cignoni et al. could be performed in approximately fifteen minutes. other more common means of 3D surface digitisation including struc- dered with greater consistency (Fig. 2008)).01) but not between the scanning Anova was used to assess the homogeneity of the results obtained techniques (p = 0. but the dif- as factors).984) even if the All the analyses were performed using the Geomorph (Adams and shape differences are localised in different areas of the cranium Otárola-Castillo.58% and 29. 2). 4). and can.only requiring a standard digital camera and freely stop of its turn-table with its internal camera and projector light and distributed software (except Geomagic that can be replaced by other this tends to produce soft contrast. max: 0. zooarchaeological specimens..088 mm (min: 0.respectively be transported (e. .21% of the total variance. Detailed views of 3D models of specimen MNHN-ZM-MO-1997-452 obtained with the Breuckmann structured light scanner (left) and photogrammetry (right). Falkingham. ilarity in the geometry of the models. photogrammetric models are produced with con. Katz and Friess.01). ble. 2010.17 mm).the two acquisition trative obstacles and time involved in processing loan permissions).04%. size. protocols for data acquisition are available and discussed in the litera- stant illumination and reconstruction algorithms that preserve a very ture (Bates et al. / Journal of Archaeological Science: Reports 9 (2016) 87–93 Anovas were also performed to test differences between the scanning Breuckmann scanner (Fig. 2014).

though potentially significant. how. In contrast. which requires only very limited ferences were observed depending of the type of landmark analysed manual cleaning and processing and no scaling before being incorporat. cal details – e. microscibe comparisons) have found relatively good congruence be- ever. landscapes or ally accurate results as good as the highest-end surface scanners artifacts. cost.47 0. We promising tool for geometric morphometric (GM) studies. exploring the possibility of combining 3D coordinates scannographic models. relatively fast and easy-to-operate acquisition protocol . Ultimately.g.g.31% of total variation). but only in with coordinates measured on 3D models obtained from various a limited set of areas displaying particularly fine and intricate topologi- sources (e.57 One should consider all trade-offs between both technologies. Top: Two first axes of the PCA showing variation between scanning technic (left) and specimens (right). acquisition time. surements that are directly comparable to those performed on the Several studies. 4. However. such as the tooth row. are largely compensated for by the benefits pho.used with free- Photogrammetry is commonly used in archaeology to construct 3D ly distributed software in default mode.49 0. can offer geometric and textur- models of architectural structures. yielding mea- certainty shown between the two techniques (6. 2011 for CT-Scans/ ences. The surface scanning time includes the 3D reconstruction.g. and results in a single.07 0.11 0.17 0. 5. we consider these drawbacks. 2011) and require care- several additional steps. involving substantial computation time and ful examination of the potential bias prior to any geometric morpho- heavier manual intervention before the raw images are turned into an metric analyses. Bottom: Shape differences along the first (left) and second (right) PCA axes. obtained through a archaeological sciences. All studies combining such data ed in the analysis. operable 3D model. . and the experience of the operator. Sholts et al. Conclusion togrammetry offers in terms of ease of use. laser or structured light scanners) (e. the ones in the negative side are in grey. 2007. which were not the primary focus of our study and do not Fig. / Journal of Archaeological Science: Reports 9 (2016) 87–93 91 Table 1 Quantitative comparison of the 3D models obtained with photogrammetry and the Breuckmann scanner. 1997–452 1997–453 1997–454 1996–2499 1997–12 Average distance 0. archaeological sites. A. .05 0. the photogrammetric procedure includes should be performed with caution (Sholts et al.with reasonable un. Specimens localised at the positive side of the axis are in black. Evin et al.. measured directly on specimens (using for example a microscribe) We found the surface scanner retained some advantages.although some dif- mesh (especially along the tooth row).61 0. The measurements are in millimeters. CT-Scans. these minimal differ- Badawi-Fayad and Cabanis. relatively more detailed good to enable geometric morphometric analyses. In this study we demonstrate that photogrammetry is also a currently used in cultural and natural heritage research facilities. tween coordinates obtained from various sources that were sufficiently lation and fusion steps.04 Standard deviation 0.. triangu. revealing demonstrate how closely these photogrammetric models match the that GM variation between specimens greatly exceeded the variation original object and argue that such an approach is ideally suited to run- produced by the other scanning technology used .52 0. ning landmark based geometric morphometric analyses. acces- sibility and maneuverability in most situations pertaining to the field of Results show that photogrammetric 3D models.

161. Tsioukas. Arctic Canada. Aside from the cost. Pastor. 48. Sci. J. Bakhshaliyev.. J. C. Hutchinson. http://dx. 500–510. A. http://dx. 18–33.) during Roman Times in Southern France. 1999.09. Larson. C. Sci.doi.. acces..005. R. I.1111/2041-210X.1169. Mach. Flink. V. M. On the trail of Neolithic mice and men towards J.doi. T.. One 8. allowing e.. Ivorra.doi. Conf.) in life: 5000 years of microevolution in the orkney vole (Microtus arvalis orcadensis)... Archaeol. 568–575. and access to the morphometric Academic.T. Sheridan. U. Pa. Sauvage. of olive stones at Ugarit. L. Quéré. Accurate measurement with photogrammetry at large sites. Pastor.1016/j. Sci.doi. 192–199. org/10.1038/srep08299. Linn.06.2007. Sapirstein.005. pp. Archaeol. Scratching the surface? The use of surface scanning in physical and paleo- in terms of the long-term curation and protection of important museum anthropology. J..1111/arcm. Geomorph: an bScp N Rb/Scp N Package for the col. 8. http://dx.G. Terral. J.doi..002.. 40. http://dx.. and robust multiview stereopsis.1016/j. C..C. Syria-a window on the Mediterranean Marinval. G. Int.. Comparison of 3D landmark and 3D dense cloud ap- Raphael Cornette and Julien Claude for the fruitful discussions we had proaches to Hominin mandible morphometrics using Structure-From-Motion. K.. D. J. 2010. R Core Team. 917–928. T.crpv. 35..1111/bij. Finally and perhaps most important Friess. T.. G. C... 1981. 290.doi. Gill.. Hist.B.L. C. 32. aqy. 2013. In: provided access to the specimens. 45. Transcaucasia: zooarchaeological clues from Nakhchivan (Azerbaijan).2008. 137–145. Rep... Cardini.. Ros. Colombeau. T. Methods Ecol. Archaeol.T. R. PLoS 567–575. M. 73–98. F. Proc. (Ed. 1990. http://dx. Tougard. 2012.jas. Popovici. Three-dimensional procrustes analysis of modern Mcpherron.B. J. B. J.1371/journal. J.1007/s00334-014-0482-y.0063195. 3D morphometric analysis of fossil canid laser scanner and a 3D digitiser: implications for applications of digital skulls contradicts the suggested domestication of dogs during the late Paleolithic.. Late Bronze Age. 1999. A.G. On the application of 3-D scanning technology for the documentation and typology of lithic artifacts. Plenum. pp. (olive) domestication: evidence for shape variation Bouby. semi-landmark morphometrics implying different template designs: the theropod formation of skull shape in St Bernard dogs. Y. Falkingham. Smikt... M. Mitteroecker.. 10. A Fragmented Past: (Re) constructing M. photogrammetric techniques to data acquisition. E. Am.. 23. Landmark methods for forms without landmarks: morphometrics of Newton... On the origins and group differences in outline shape. Bouby. Sholts.. T. Evin. A. Adams. 19–24. P. 2014. and colour of the objects in a more useful manner..2012. http://dx. (avail. J. Seventh IEEE 393–399. MeshLab: an open-source 3D mesh processing comparative shape analysis of archaeological and modern seeds. IEEE Trans. Dobney. Gunz. light surface scanning with a relatively short time needed for data acqui. J.. A.2013.. Archaeol. 015. Katz. K.. Bates. P.. 75–84. 80–89. Sci.2015. Biol. Rmorph: a R geometric and multivariate morphometrics library. Lacombe. Adv. T. Sci.. Wickham-Jones. 11–22..07.doi. Hodgetts. of Leeds Press. Comparison of coordinate leobiology 7. 2014. morphometrics.2013. http://dx. A biometric re-evaluation of recent claims for 4420–4428.. http://dx. 181–204. Med. Langdon. Coquerelle. A. Methods for Antiquity through 3D Artefact Modelling and Customised Structured Light Scanning 3D digitization of Cultural Heritage. (Eds.0616. Hulme-Beaman... crania-a preliminary assessment. Bailey. http://dx. G. southern France. Pascal. Antiquity 90. 24.. could be easily resolved with minor Evin. Anat.doi.. http://dx. S. Lewis-Bale.G. A.. Grosman. Seetah. lection and analysis of geometric morphometric shape data..pone.. 40. A. S. C. J.1016/j. photogrammetry also captures the texture Trans. Goodall. Paris. Biological Structure and Images. Corsini. Picq. Technical note: 3D from standard digital photography of human prove the manuscript.. Walcek. (grant number NE/K003259/1) and the European Research Council Haukaas. This is publication ISEM 2016-127. Pattern Anal. Mirea..J. Extensions of the procrustes method for the optimal superimpo- Cucchi.doi. Rec. Herit. Smilansky..06. Evin. 2014. F. Flores. 206–218.. Hystrix 24. Larson. M. L.doi.2013. T. Ivorra.2006.2008. Figueiral. Slice. A. M. 2013. Intell. A. Renaud. Terral. J. S.. Souter. Berthon. 7–31.1016/ at Athienou-Malloura. P. tial-temporal variation ? Population dynamics of Ursus spelaeus from Romualdova pe Davis. Cucchi..doi.G. D..doi.L. L. 2016. C. Andreescu. C. 1150–1157. R: A Language and Environment for Statistical Computing.4404/hystrix-24. A.. D.. 1997. 36. Morey. Rovira.. Geometric morphometric analysis of grain shape Heckel. Structured light scanning for high-resolution human craniofacial form. Walker.90004. 93–98. (grapevine) in ancient Greece by Cignoni. P. I. J... Inf. Bioarchaeological insights into the process of domesti.02. 2.. Syst.doi. Garstki.. Siahsarvie. Z.doi. J. 2016. Current Issues in Statistical Shape Analysis. Pascal. Accurate. L. J. http://dx..J.R. 137–139.1016/j. Roussel.181. 2003. S. Philos. Tresset. Evolution (N.1098/rstb. Bouchette. A simple photogrammetry rig for the reliable Cucchi.12229.2014. documentation of in situ archaeological finds. D.1016/0305-4403(92)90049-9. S. Rarity. 1362–1376. L. shape. P.. Buill. 2804–2820. Photogrammetry offers a very low cost alternative to structured Evin. O'Higgins.. 2013... Ruas. Petit.028. P.. 66.-P. Evin et al. http://dx. http://dx. http://dx.M. dense. A. G. P. 41.R.07. of Late Pleistocene to Holocene mammals of Israel the effects of temperature change jas. 411–422.E. 2014.. R. 2009. Application of high-resolution laser scanning and doi. Proc. pelvis and the shrew skull. T... S.doi. Porter.J. Acquisition of high resolution three-dimensional models using free. Archaeol.. Dobney. 2011.doi. Escarguel. R..jas.. A... Bookstein. 2007. J. Sci.. Terral. T. M. F.B. Semilandmarks: a method for quantifying curves and sur- faces.1016/j. Unravelling the complexity of domestication: a case study using mor- sition.. Sto- chastic Geometry. Chamzas. Averett P.. jas.jas..1016/j.. This work was supported by NERC Archaeometry http://dx.006. 370. 71–86..A. T.. S. The effects of temperature change and domestication on the body size cina and U. Boronean. 3. . Rohlf.. K.. 19.011. 3D model of the Can Sadurní cave.1016/j.. 2016. Lorre. Mainland.. Bookstein. A. Bonsall.C. Sci. J. Cornette. P. Lowe. 108. P..jas. 71–76. K..790410. G. 2011. sition of landmarks. Dobney. A. Searle.. Marro. http://dx.doi.F. K. http://dx. 1992...jas. Anthropol. 2008.doi.. F. Archaeol. system. 21. New platform where the surface scans were performed. 2014. Int. when writing the manuscript. 2014.doi. Sci. In: J. points to be more easily observed..024..F. Leeds. Evol. 4. F.M. Multivariate regression models and geo-metric morphometrics: the palaeontology. 38.007. Blescu..jas. Pedraza.. Modern Morphometrics in Physical Anthropology. 1–8. Ivorra.2011. shape and development in the evolution of the domestic dog. Badawi-Fayad..K. New-York. Procrustes methods in the statistical analysis of shape revisited.A.doi.1111/evo. (Croatia). Hublin. Archaeol. 40–56. U.P... S.1109/ICCV. (Eds. Kent. Inferring the agrobiodiversity of Vitis vinifera L.M. Archaeol..... D. 15 (1T:15p). http://dx. Comptes Rendus Palevol 9.. The untapped potential of low-cost photogrammetry in community-based archaeology: a case study from Banks Island. Klingenberg. Cucchi. spread of Olea europaea L. ERCIM News 2008. Cult. http://dx. T. K. 39. M... morphometric approaches. XXXVIII 68–73.L.. Brehard. Early Upper Palaeolithic wolf domestication in Eurasia.1-6292. 39. L. P..10. C. 4. http://dx.12004. Orth. 275. Spat. L. Cucchi. We also thank York.10. 225–243. Dobney.. 2012. The long and winding road: identifying pig domestication through molar size and tuning of the photogrammetric protocol.08. Mardia. Baylac. A. Sci.. S. G. Two applications of 3D Drake. D.. J. Biol. K. that Gunz.V. Renvoisé... Picq.. We thank the two anonymous reviewers who greatly helped to im.09. http://dx. Evin.. Comput. Koutsoudis.R. Yuan.P... Object recognition from local scale-invariant features. Anthropol. 2012.. Soressi. Mitteroecker.1098/rspb.doi.. J. Bouby. 268–276.028. J.. E. 535–543. M. http://dx. Community Archaeol. photogrammetry will provide key accurate replica archives Furukawa. Sci. B Biol. Baylac. J. F.. Principal component analysis for shape variation about an References underlying symmetric shape. 2014. R... photogrammetric software. D. Arch. Vidarsdottir. C. http://dx.15184/ j..T. Pract. 1. In: Supplementary data to this article can be found online at http://dx.. We thank the Muséum National d'Histoire Naturelle. M. Soc. Edo. Photogramm. collections.2010. A.. Hist. Herit. 1995.S.T. Mardia. 90.2010. cation of grapevine (Vitis vinifera L. Boudadi-Maligne..T. J. Image Anal. Veg.1016/j. analysis and interpretation in Monteiro. Henan Province.. http://dx. S. 40–59. Ranzuglia.. Hassett. Palaeontol. able on request at baylac@mnhn. Biol. 2013.. M. doi.. e63195.. T.L.. C.. J.1999. Osteoarchaeol. (ERC-2013-StG 337574-UNDEAD).. Friess.2009. Otárola-Castillo. Cyprus.). 3101–3110.M. The pace of morphological change: historical trans.F. Can morphospace shed light on cave bear spa- Soc. Wärmländer. A. C.12476. Sci.22468... Electron. J. Leeds University Press. Darvish.. S. Y) 68. Miracle. The changing pace of insular and the identification of two-rowed barley (Hordeum vulgare subsp.. 2005.2016. 2010.1109/TPAMI.culher..006. Remote Sens.. Strand Vidarsdottir. open-source.. J. 2016... that can be shared and made more widely accessible for research. / Journal of Archaeological Science: Reports 9 (2016) 87–93 affect the morphometric results. K... 101–114. O.. search for causal factors in the analysis of shape.. C. M.. S.. Núñez. 2008. sibility and time efficiency. Phys. Lazr.1016/j. M. Kovács.doi. D. Early Neolithic pig domestication creation of 3D artifact models in the field lithic examples from the early upper Pa- at Jiahu. 2013. R. V. Archaeobot. Sci. Martínková. and using relatively basic equipment. Sci. P. 2013. Archaeol. Archaeol.). J. Basin and on the westward diffusion of olive varieties.V.P. M. Pagnoux. S.4436/jass. R.. 2014. Vis. N.doi. Acknowledgements http://dx.92 A.1016/j. Semilandmarks in three dimensions. Kolamunnage. Veg. distichum L. P.K.1002/ajpa. Leeds. E. Valamoti. China: clues from molar shape analyses using geometric leolithic sequence of Les Cottés (France).. measurement precision of different landmark types on human crania using a 3D Barnett. 152–158. Sci. 2016.06.doi. suture Falkingham.). 20130616. http://dx. Biol. Cardini. C..jasrep. Cabanis. Bonhomme. K. University doi. 735–743.. Sci.-F. pp. L.doi.G. phometrics and ancient DNA analyses of archaeological pigs from Romania. F. and domestication on the body size of Late Pleistocene to Holocene mammals of. J.

This.. Torres. Towards Linear-Time Incremental Structure from Motion. Pastor. 2012... Neiswanger. G. 2007. Vermeulen... Lacombe. S. SiftGPU : A GPU Implemention of Scale Invariant Feature Transform (SIFT) aob/mcp298. C. International Conference on 3D S.. Curless. VisualSFM: A visual structure from motion system [WWW metric perspectives to understand seed domestication syndrome and reveal origins document]http://ccwu. construct underwater shipwreck sites. rial photographs. P. Chevance. J. B.M.022.. Bot. S.5995552. F.1016/0278-4165(91)90021-O. Ivorra. Tardy. J. Face shape of unaffected parents with cleft affected offspring: combining three- View publication stats .1111/j.. R. 3057–3064 http://dx.. 2016.25 IEEE. 105. R. Picq. lution and history of grapevine (Vitis vinifera) under domestication: new morpho. Briese. 54–75. Naidoo. Ann. velopment on morphological change..01462. J. Vis. http://dx. Yamafune. Evin et al. 443–455. 2013 Terral. Verhoeven. Doneus..2009. http://dx. 1991. E. I. Y) 40.x. Multi-image photogrammetry to record and re- Wayne.doi.doi. 02. dimensional surface imaging and geometric morphometrics. 2060–2070. 2011 IEEE Conf. 39.1109/3DV. Anthropol. http:// 12.doi. A. S. 271– C. 2009.Face... M.. T.. Multicore bundle adjustment. C. Orthod Carniofac Res. Resick.1007/s10816-016-9283-1. 10. 2013.M.. K. C.. 2011.. Archaeol. of ancient European cultivars.. C. K. Seitz. scious selection in primeval domesticates. M. C. 2013. pp. L. 1986.D.... [WWW Document].2011. Castro. Jung. 2011. Archaeol.jas. J.-B. Cranial morphology of domestic and wild canids : the influence of de. Tabard. (CVPR). F.doi. E. Wu. Bacilieri.1016/j. L.. C. Mapping by matching: a com. Bardi. / Journal of Archaeological Science: Reports 9 (2016) 87–93 93 Tchernov. Evolution (N.M. puter vision-based approach to fast and accurate georeferencing of archaeological ae. Figueiral. Bouby. T.doi. Sci. R. org/10.. http://dx. Evo. Wu... Compan.. Pattern Recognit.. S.. K.doi... 243–261.... J. Method Theory 1–23 http://dx. Kolska Horwitz.M. 127–134 http://dx. dx. C. 2010. Archaeol. Body size diminution under domestication: uncon.1093/ Wu. Agarwal. S.. Brandon.. L. Fabre. 1109/CVPR. J.