WSCG 2014 - repositorium.sdum.uminho.pt

22nd

International Conference in Central Europe

on

Computer Graphics, Visualization and Computer Vision

in co-operation with

EUROGRAPHICS Association

WSCG 2014

Full Papers Proceedings

Edited by

Vaclav Skala, University of West Bohemia, Czech Republic

22nd

International Conference in Central Europe

on

Computer Graphics, Visualization and Computer Vision

in co-operation with

EUROGRAPHICS Association

WSCG 2014


Edited by

Vaclav Skala, University of West Bohemia, Czech Republic

Vaclav Skala – Union Agency

WSCG 2014 – Full Papers Proceedings

Editor: Vaclav Skala

c/o University of West Bohemia, Univerzitni 8

CZ 306 14 Plzen

Czech Republic

[email protected] http://www.VaclavSkala.eu

Managing Editor: Vaclav Skala

Published and printed by:

Vaclav Skala – Union Agency

Na Mazinách 9

CZ 322 00 Plzen

Czech Republic http://www.UnionAgency.eu

Hardcopy: ISBN 978-80-86943-70-1

mailto:[email protected]

http://www.vaclavskala.eu/

http://www.unionagency.eu/

WSCG 2014

International Program Committee

Balcisoy,Selim (Turkey)

Benes,Bedrich (United States)

Benger,Werner (United States)

Bittner,Jiri (Czech Republic)

Daniel,Marc (France)

Daniels,Karen (United States)

de Geus,Klaus (Brazil)

Debelov,Victor (Russia)

Feito,Francisco (Spain)

Ferguson,Stuart (United Kingdom)

Flaquer,Juan (Spain)

Gain,James (South Africa)

Gavrilova,M. (Canada)

Giannini,Franca (Italy)

Guthe,Michael (Germany)

Herout,Adam (Czech Republic)

Choi,Sunghee (Korea)

Juan,M.-Carmen (Spain)

Kalra,Prem K. (India)

Kellomaki,Timo (Finland)

Kim,HyungSeok (Korea)

Klosowski,James (United States)

Kurt,Murat (Turkey)

Max,Nelson (United States)

Muller,Heinrich (Germany)

Murtagh,Fionn (United Kingdom)

Oliveira,Manuel M. (Brazil)

Oyarzun Laura,Cristina (Germany)

Pan,Rongjiang (China)

Patow,Gustavo (Spain)

Platis,Nikos (Greece)

Puppo,Enrico (Italy)

Renaud,Christophe (France)

Richardson,John (United States)

Rojas-Sola,Jose Ignacio (Spain)

Sabine,Coquillart (France)

Santos,Luis Paulo (Portugal)

Segura,Rafael (Spain)

Semwal,Sudhanshu (United States)

Sousa,A.Augusto (Portugal)

Stroud,Ian (Switzerland)

Szecsi,Laszlo (Hungary)

Teschner,Matthias (Germany)

Theussl,Thomas (Saudi Arabia)

Tokuta,Alade (United States)

Wuensche,Burkhard,C. (New Zealand)

Wuethrich,Charles (Germany)

Zemcik,Pavel (Czech Republic)

Zitova,Barbara (Czech Republic)

Zwettler,Gerald (Austria)

WSCG 2014

Board of Reviewers

Abad,Francisco

Adzhiev,Valery

Agathos,Alexander

Alvarado,Adriana

Assarsson,Ulf

Ayala,Dolors

Backfrieder,Werner

Balcisoy,Selim

Barbosa,Joao

Barthe,Loic

Battiato,Sebastiano

Benes,Bedrich

Benger,Werner

Bilbao,Javier,J.

Billeter,Markus

Biri,Venceslas

Birra,Fernando

Bittner,Jiri

Bosch,Carles

Bourdin,Jean-Jacques

Brun,Anders

Bruni,Vittoria

Buehler,Katja

Bulo,Samuel Rota

Cakmak,Hueseyin Kemal

Camahort,Emilio

Casciola,Giulio

Cline,David

Coquillart,Sabine

Cosker,Darren

Daniel,Marc

Daniels,Karen

Debelov,Victor

Drechsler,Klaus

Durikovic,Roman

Eisemann,Martin

Erbacher,Robert

Feito,Francisco

Ferguson,Stuart

Fernandes,Antonio

Flaquer,Juan

Fuenfzig,Christoph

Gain,James

Galo,Mauricio

Garcia Hernandez,Ruben Jesus

Garcia-Alonso,Alejandro

Gavrilova,M.

Geus,Klaus de

Giannini,Franca

Gobron,Stephane

Gobron,Stephane

Gois,Joao Paulo

Grau,Sergi

Gudukbay,Ugur

Guthe,Michael

Haberdar,Hakan

Hansford,Dianne

Haro,Antonio

Hasler,Nils

Hast,Anders

Hernandez,Benjamin

Herout,Adam

Herrera,Tomas Lay

Hicks,Yulia

Hildenbrand,Dietmar

Hinkenjann,Andre

Horain,Patrick

Chaine,Raphaelle

Choi,Sunghee

Chover,Miguel

Chrysanthou,Yiorgos

Chuang,Yung-Yu

Iglesias,Jose A.

Ihrke,Ivo

Iwasaki,Kei

Jato,Oliver

Jeschke,Stefan

Joan-Arinyo,Robert

Jones,Mark

Juan,M.-Carmen

Kalra,Prem K.

Kämpe,Viktor

Kanai,Takashi

Kellomaki,Timo

Kim,HyungSeok

Klosowski,James

Kolcun,Alexej

Krivanek,Jaroslav

Kurillo,Gregorij

Kurt,Murat

Kyratzi,Sofia

Larboulette,Caroline

Lee,Jong Kwan

Liu,Damon Shing-Min

Lopes,Adriano

Loscos,Celine

Lutteroth,Christof

Maciel,Anderson

Mandl,Thomas

Manzke,Michael

Marras,Stefano

Masia,Belen

Masood,Syed Zain

Masso,Jose Pascual Molina

Matey,Luis

Max,Nelson

Melendez,Francho

Meng,Weiliang

Mestre,Daniel,R.

Meyer,Alexandre

Molla,Ramon

Montrucchio,Bartolomeo

Morigi,Serena

Mukai,Tomohiko

Muller,Heinrich

Munoz,Adolfo

Murtagh,Fionn

Okabe,Makoto

Oliveira,Joao

Oliveira,Manuel M.

Oyarzun,Cristina Laura

Pan,Rongjiang

Papaioannou,Georgios

Paquette,Eric

Pasko,Galina

Patane,Giuseppe

Patow,Gustavo

Pedrini,Helio

Pereira,Joao Madeiras

Peters,Jorg

Pina,Jose Luis

Platis,Nikos

Post,Frits,H.

Puig,Anna

Puppo,Enrico

Puppo,Enrico

Rafferty,Karen

Raffin,Romain

Renaud,Christophe

Reshetouski,Ilya

Reshetov,Alexander

Ribardiere,Mickael

Ribeiro,Roberto

Richardson,John

Ritter,Marcel

Rojas-Sola,Jose Ignacio

Rokita,Przemyslaw

Rudomin,Isaac

Sacco,Marco

Sadlo,Filip

Salvetti,Ovidio

Sanna,Andrea

Santos,Luis Paulo

Sapidis,Nickolas,S.

Savchenko,Vladimir

Segura,Rafael

Seipel,Stefan

Sellent,Anita

Semwal,Sudhanshu

Sheng,Bin

Sheng,Yu

Shesh,Amit

Schmidt,Johanna

Sik-Lanyi,Cecilia

Sintorn,Erik

Sirakov,Nikolay Metodiev

Sourin,Alexei

Sousa,A.Augusto

Sramek,Milos

Stroud,Ian

Subsol,Gerard

Sundstedt,Veronica

Svoboda,Tomas

Szecsi,Laszlo

Tang,Min

Teschner,Matthias

Theussl,Thomas

Tian,Feng

Tokuta,Alade

Torrens,Francisco

Trapp,Matthias

Tytkowski,Krzysztof

Umlauf,Georg

Vergeest,Joris

Vitulano,Domenico

Vosinakis,Spyros

Walczak,Krzysztof

Wan,Liang

Wu,Shin-Ting

Wuensche,Burkhard,C.

Wuethrich,Charles

Xin,Shi-Qing

Xu,Fei Dai Dongrong

Yoshizawa,Shin

Yue,Yonghao

Yue,Yonghao

Zalik,Borut

Zemcik,Pavel

Zhang,Xinyu

Zhao,Qiang

Zheng,Youyi

Zitova,Barbara

Zwettler,Gerald

WSCG 2014


Contents

Keynote Talks

Weinkauf,T.: Flow maps - Benefits, Problems, Future Research i

Oliveira,M.M.: Performing High-Dimensional Filtering in Low-Dimensional Spaces ii

Barsky,A.: Simulating Human Vision and Correcting Visual Aberrations with

Computational Light Field Displays

iii

Papers Page

Khatri,N., Joshi,M.V.: Efficient Self-learning for Single Image Upsampling 1

Brix,T., Lindemann,F., Prassni,J.-S., Diepenbrock,S., Hinrichs,K.: Visual Analysis

of Polarization Domains in Barium Titanate during Phase Transitions

9

Falschlunger,L., Eisl,C., Losbichler,H., Greil,A.: Improving Information Perception

of Graphical Displays: an Experimental Study on the Display of Column

Graphs

19

Lee,J.H., Kim,M.H.: Locally Adaptive Products for Genuine Spherical Harmonic

Lighting

27

Russell,A., Kamayou,F., Marceau,R., Daniels,K., Grinstein,G.: Point Sensitivity

for Radial Visualization under Dimensional Anchor Motion

37

Jozsa,P., Toth,M.J., Csebfalvi,B.: Analytic Isosurface Rendering and Maximum

Intensity Projection on the GPU

47

Szécsi,L., Szirányi,M.: Recursive Procedural Tonal Art Maps 57

Kuri,D., Root,E, Theisel,H.: Hexagonal Image Quilting for Texture Synthesis 67

Chajdas,M. G., Reitinger,M., Westermann,R.: Scalable rendering for very large

meshes

77

Oh, S., Jeong, I.K.: Single-phase trapped air simulation in water flow 87

Kajan,R., Szentandrási,I., Herout,A., Pavelková,A.: Reliable and Unobtrusive

Inter-Device Collaboration by Continuous Interaction

95

Graszka,P.: Median mixture model for background - foreground segmentation in

video sequences

103

Mousas,C., Newbury,P., Anagnostopoulos,C.N.: Efficient Hand-Over Motion

Reconstruction

111

Kato,T., Saito,S., Kawai,M., Iwao,T., Maejima,A., Morishima,S.: Character

Transfer: Example-based individuality retargeting for facial animations

121

Yusuf,A., Mostafa,M., Elarif,T.: A Multi-threaded Multi-context GPU Methodology

for Real-time Terrain Rendering

131

Du,S., Kanai,T.: GPU-based Adaptive Surface Reconstruction for Real-time SPH

Fluids

141

Lemasson,G., Lehl,J.C., Zara,F., Baudet,V., Arthaud,P., Shariat,B.: Accurate

thickness computing of a B-Rep model on the GPU

151

Bernardes,P., Magalhaes,F., Ribeiro,J., Madeira,J., Martins,M.: Image-based 3D

Modelling in Archaeology: Application and Evaluation

159

Walek,P., Jan,J., Ourednicek,P., Skotakova,J., Jira,I.: Improved Estimation of

Tissue Noise Power Spectra in CT Data

167

Mujika,A., de Mauro,A., Robin,G., Epelde,G., Oyarzun,D.: A Physically-based

Simulation of a Caenorhabditis elegans

177

Jribi,M., Ghorbel,F.: A Stable and Invariant Three-polar Surface Representation:

Application to 3D Face Description

185

Debiasi,A., Simoes,B., De Amicis,R.: Supervised Force Directed Algorithm for the

Generation of Flow Maps

193

Santos,A.L., Teichrieb,V., Lindoso,J.E.F.: Review and Comparative Study of Ray

Traversal Algorithms on a Modern GPU Architecture

203

Marcovecchio,D., Delrieux,C., Werdin,J., Stefanazzi,N.: A Lightweight Approach

for Analyzing Insect Behavior on a Mobile System

213

Image-based 3D Modelling in Archaeology: Application andEvaluation

Paulo BernardesArchaeology Unit,

University of MinhoAvenida Central, 394710-278, Braga,

[email protected]

Fernanda MagalhãesArchaeology Unit,


[email protected]

Jorge RibeiroArchaeology Unit,


[email protected]

Joaquim MadeiraDETI/IEETA, University of

AveiroCampus Universitário de

Santiago3810-193, Aveiro,

[email protected]

Manuela MartinsArchaeology Unit,


[email protected]

ABSTRACTImage-based 3D modelling tools and techniques can be used to support some stages of the archaeological process.Two application examples, for two sites of the Roman Era, are presented, illustrating the usage and usefulnessof such tools for the archaeological survey. For a quadrangular pool in the Milreu Roman Villa (Faro, Portugal),the 3D models resulting from the application of two different image-based modelling tools, using the same initialset of digital photos, are compared regarding time spent and model accuracy. For the Fonte do Ídolo in BracaraAugusta (Braga, Portugal), the result of a traditional survey is compared both with a laser-based survey and animage-based survey.

KeywordsImage-based modelling, Archaeological process.

1 INTRODUCTION

As is well-known, the archaeological process comprisesseveral steps that begin with the excavation and ar-chaeological survey, proceed to the analysis and inter-pretation stage, and end with the dissemination of re-sults. Archaeological interpretation and research de-pend upon accurate data collection during excavation.Traditional data collection, as stated in [DeR13a], is al-ways a two-dimensional survey of a three-dimensionalreality, which certainly introduces potential inaccura-cies. An excavation is always a destruction, but thepurpose of archaeological survey is, exactly, to mini-

Permission to make digital or hard copies of all or part ofthis work for personal or classroom use is granted withoutfee provided that copies are not made or distributed for profitor commercial advantage and that copies bear this notice andthe full citation on the first page. To copy otherwise, or re-publish, to post on servers or to redistribute to lists, requiresprior specific permission and/or a fee.

mize the consequences of this irreversibility. Hence,a major concern of different archaeology teams in anarchaeological site should always be to ensure the vi-sual accuracy and the exactitude of the archaeologicalrecord.

The quality of the archaeological record regarding ac-curacy is obviously a great concern of all archaeologyteams. In terms of accuracy, archaeological data acqui-sition using laser technology is certainly a valid deci-sion. However, the high cost of the devices, as well asthe severe conditions of some archaeological sites donot facilitate its use. Therefore, the use of image-basedtechniques, based on structure from motion algorithmsfrom computer vision, to automatically perform 3D re-construction, appears to be a valid and affordable optionthat has increasingly gained adherents in archaeology[Bru12a].

Although image-based modelling techniques have onlylately gained some importance in archaeology, their usein heritage is not recent. Thus, the following sectionwill be a brief retrospective of the usage of this technol-

WSCG 2014 Conference on Computer Graphics, Visualization and Computer Vision

Full Papers Proceedings 159 ISBN 978-80-86943-70-1

ogy, trying to analyse which stages of the archaeologi-cal process have most taken advantage of this practice.

In the third section two examples will be presented,where image-based modelling techniques were used forarchaeological data acquisition, carried out within dif-ferent projects of the Archaeology Unit of the Univer-sity of Minho (UAUM). The first example allows com-paring the performance of two free image-based mod-elling packages, regarding mesh density and time nec-essary to produce a 3D model. The 3D reconstructionswere done using the same dataset of a quadrangularpool located at the Roman Villa in Milreu (Faro, Por-tugal). The other example concerns the archaeologicalrecord of a monument in Braga (Portugal) known asFonte do Ídolo. Three distinct data acquisition proce-dures were carried out: the first took place in 2005 andwas performed according to the traditional methodol-ogy, the second was perfomed in 2008 and used a to-tal station and a hand-held laser scanner, the latter wascompleted recently based on simple photos. The threemethodologies are compared regarding the time usedfor each data acquisition and the quality of the acquireddata.

The last section is reserved for the presentation of someconclusions regarding the usage of image-based mod-elling during different stages of the archaeological pro-cess and to point out some guidelines/ideas that mightcontribute to accomplish the needs of archaeologistsduring the entire archaeological process.

2 IMAGE-BASED MODELLING INHERITAGE

Already in 2000, the 3D MURALE project developedimage-based 3D capture and visualisation technologiesfor archaeology, using as test case the ancient city ofSagalassos (Turkey) [Pol01a]. The main goal of thatEuropean IST project was to develop affordable toolsthat would be easy to use by archaeologists to presentdifferent types of archaeological objects with accept-able visual quality.

Later, in 2003, the EPOCH network gathered severalEuropean cultural institutions to join efforts for im-proving the quality and effectiveness of the use of in-formation and communication technologies for culturalheritage [Arn07a]. Within this project and regardingimage-based modelling, the Vision for Industry Com-munications and Services group at the Katholieke Uni-versiteit Leuven (Belgium) developed a web-based 3Dreconstruction service to be used in the cultural heritagefield. This service, called ARC 3D, starts with the up-loading of digital images of an object or scene. Theautomatic reconstruction process computes the cameracalibration, as well as dense depth maps for the images.The reconstruction result can be later downloaded andvisualized on a workstation. Also under this project, the

Visual Computing Lab at ISTI-CNR (Italy) developedan extremely useful and open-source tool, called Mesh-Lab, for processing and editing 3D triangular meshes[Cig08a].More recently, the 7th Framework Programme fundedthe 3D-COFORM project, that brought together notonly several former EPOCH partners but also otherprestigious European cultural heritage organizations.The aim of this consortium was to promote and en-hance the use of 3D data for long term documentationof tangible cultural heritage. To achieve this the 3D-COFORM consortium sponsored several research ac-tivities that established an important set of tools to beused in different scenarios such as archaeological andhistoric urban site modelling or 3D artefact acquisition.But the utility of these tools is only recognised if theyare successfully used, therefore 3D-COFORM estab-lished a Virtual Centre of Competence in 3D to promotetheir role in cultural heritage projects [Nic10a].As a matter of fact, there are several archaeol-ogy projects that are actively using image-basedmodelling techniques for architectural structuresacquisition [San13a] or during the excavationprocess [Her12a][DeR14a][McC14a]. There areseveral different image-based modelling packagesavailable, but according to the surveyed literaturethe more common are 123DCatch from Autodesk[Bru12a][San13a][McC14a] and PhotoScan fromAgisoft [Bru12a][DeR13a]. Alternatively, opensource packages such as VisualSfM may also be used[DeR13a][McC14a].

3 USING FREE IMAGE-BASED MOD-ELLING TOOLS

As stated in [DeR13a], the 2D representation of 3D ar-chaeological entities generates some loss of informa-tion. Therefore, various computer vision techniques,such as structure from motion (SfM) and dense stereo-reconstruction algorithms, are already used for recover-ing 3D shape and appearance of archaeological objectsin some archaeological excavations [Her12a]. Besidesthe advantages for the archaeological record, it enablesalso the use of real 3D data in the visualization process.There are several low-cost or free computer visionbased software packages that enable the generation of3D point clouds and the representation of 3D meshes.Since some archaeology projects deal with severebudget constrains, two free software packages wereselected for comparison: (1) 123DCatch from Autodesk[123D] and (2) VisualSFM [Wu13]. While the twopackages were used and compared in the case of theRoman pool in Milreu, for the reconstruction of theFonte do Ídolo only 123DCatch was used.123DCatch is extremely straightforward and easy touse. It takes only a few simple steps to create a real-



istic 3D representation of an archaeological object. Thefirst step is to take digital photos of the object that hasto be represented. It is recommended to shoot at leastone loop of about 20 sequential photos in small incre-ments. If it is physically possible, another loop froma different angle should be taken, since this will im-prove the quality of the reconstructed 3D mesh. Also,to improve reconstruction results sticky targets shouldbe placed around the object. The following step uploadsthe photos to the cloud, where the 3D model is createdand saved into the cloud storage space. The creation ofthe 3D model is not dependent on any user configura-tion. Finally, some post-processing has to be performedon the 3D model, in order to isolate the data that is re-ally necessary for archaeological research.

VisualSFM is an open-source graphical user interfaceapplication for 3D reconstruction using SfM. As in theprevious software package, the first step is to take dig-ital photos. However, unlike 123DCatch these photosare not uploaded into the cloud, but they are added intothe VisualSFM workspace. The next step is to performfeature detection and full pairwise image matching. Atany time it is possible to add more photos and to per-form a customized feature detection and image match-ing. After the matching stage the user is able to per-form a sparse reconstruction that precedes a dense re-construction. This dense reconstruction delivers a 3Dpoint cloud that still has to be transformed into a 3Dmesh. The 3D mesh and the final post-processed recon-structed object are obtained with MeshLab.

The quadrangular pool of the Roman Villain Milreu (Faro, Portugal)The Roman ruins of Milreu are located outside Faro,near the village of Estoi. They are considered an impor-tant archaeological site in the Algarve region. The rich-ness of this villa is shown by its considerable amount ofarchaeological structures and findings [Hau02a]. Thestructure of interest is a quadrangular pool coated withmarine themed mosaics, referred as mosaic number 2and carefully described in [Car08a].

Before starting the photographic survey of the quad-rangular pool, targets were placed around the structure.Targets are usually white circles with a black circumfer-ence of 1cm in diameter and a concentric black circle of5mm in diameter, printed on adhesive labels. These tar-gets are generally used to improve the reconstruction re-sults and, in a later processing phase, for manual stitch-ing of vertices. Figure 1 shows the distribution of sometargets over the pool.

The survey was performed using a Canon EOS 550D,with which a loop of 24 photos was shot. The photoshave enough overlapping and at least three targets arealways visible. The entire 3D reconstruction processwas carried out on a ASUS K55V with an Intel Core

i5-3210M (2,5GHz) processor, 4GB memory, a 64-bitoperating system (Windows 7) and a 2GB DDR3 nVidiaGeForce GT 630M graphics card.

Figure 1: Distribution of the sticky targets over the sur-face of the quadrangular pool

Figure 2 displays the 123DCatch user interface withthe reconstruction of the quadrangular pool without anypost-processing of the mesh. It took about 27 min-utes to upload the 24 pictures and to compute a meshof 229,072 vertices and 366,570 triangles. As a matterof fact, 3D model computing rather depends on the in-ternet access upload speed, than on the computer hard-ware.

Figure 2: Pool reconstruction with 123DCatch

The same structure was also reconstructed with Visu-alSFM, using the same images and the same hardware.After adding the photos into the VisualSFM workspace,the sparse reconstruction generated a point cloud with11,304 points in 28 seconds, while the dense recon-struction generated a point cloud with 1,495,114 pointsin 10h 16m 10s. Both point clouds are represented infigure 3.

After the point cloud generation, MeshLab was usedto create a 3D mesh from the point set. This com-prises several steps: (1) computation of normals for thepoint set; (2) for a dense point cloud a point samplingwill be necessary; (3) finally, points and normals areused to build a 3D surface using the Poisson Surfacereconstruction algorithm. Figure 4 shows very satis-



Figure 3: Sparse (left) and dense (right) point cloud of the quadrangular Pool

Figure 4: 3D mesh of the Pool

factory mesh reconstruction with 833,051 vertices and1,666,062 triangles.

The Fonte do Ídolo of Bracara Augusta(Braga, Portugal)The Fonte do Ídolo, in Braga, Portugal, is a roman reli-gious stucture from the I century AD which was discov-ered in the late XVII century and has been studied sincethen. The monument has a 6m× 2.20m granite façadewith two sculptures and several epigraphs, encompass-ing an overall area of about 78m2 (13m×6m) [Ele08a](see figure 5).

Traditional Survey

In 2005 a major archaeological survey took place in or-der to properly document this archeological site. Theinterpretation of this site was based on the analysed andperceived reality, however this survey, which is a testi-mony of the site, enhanced its interpretation. The sur-vey was performed using manual drawing, describingscrupulously both details and measures of the identifiedstructures. This task was performed not only with thetraditional tape measures, graph paper and plumb line,

Figure 5: The Fonte do Ídolo

but also using topographic precision equipment, suchas a total station (Nikon Total Station DTM 310) and alevel (SOKKIA B20), to ensure the correct georeferenceof the monument within the city mesh.

All structures were carefully recorded by drawing theirplan over a graph paper. The chosen scale was 1:20 inorder to provide a detailed and precise representation.



The façade of the monument was also recorded, withthe same scale, reproducing the various figures and in-scriptions present in this granite outcrop. The drawingreplicates on paper the outlines of the overall structuresand figures. This applied methodology enabled fur-thermore the establishment of the built stratigraphy unit(UE), regarding each of the detected structures. Finally,the survey was completed with a detailed and compre-hensive photographic record.

The entire survey was carried out by two experiencedarchaeologists during two weeks. After the survey, thedrawings were scanned and vectorized with a CAD sys-tem, to be later used for research and dissemination pur-poses (see figure 6).

Figure 6: CAD processed traditional survey

Laser Scanner and Total Station based SurveyAnother survey of the Fonte do Ídolo took place in 2008and it combined data acquisition with a total station anda laser scanner. A detailed description of the 3D ac-quisition process and the purpose of this survey can befound in [Bar08a]. In this 3D geometric acquisition thetotal station (Nikon Total Station DTM 310) was used toget a geo-referenced, but less detailed, 3D mesh (15,621vertices and 29,635 triangles) of the granite monoliththat involves the monument, while the 3D laser scanner(ZScanner 700 from Z Corporation) was used to obtain

an accurate and very detailed 3D model (1,137,721 ver-tices and 2,263,090 triangles) of the sculptured façadeof the monument.

Figure 7: Combined laser and total station 3D recon-struction of the Fonte do Ídolo

The extremely dense point cloud had to be simplifiedfor manipulation purposes. Therefore, to reduce thecomplexity of the mesh the Normal-based Simplifica-tion Algorithm (NSA) described in [Fru07a] was used,which reduced the mesh complexity to approximately22% of its original size (500,000 triangles). After sim-plifying the façade of the monument, both meshes weremerged to obtain the final 3D model (figure 7). Two ex-perienced surveyors took less than a day for the overalldata acquisition.

Image-based SurveyMore recently, the UAUM carried out an image-basedsurvey of the Fonte do Ídolo for evaluation and compar-ison purposes. The entire survey was performed with49 photos recorded with a Canon EOS 550D and the3D reconstruction was carried out on a Fujitso CelsiusW520 with an Intel Xeon CPU E3-1240 V2 (3,4GHz)processor, 16GB memory, a 64-bit operating system(Windows 7) and a 1GB GDDR5 nVidia Quadro 2000Dgraphics card. It was not necessary to associate stickytargets, since the site has enough and significant naturalfeatures.The survey was carried out by an archaeologist whoparticipated in the 2005 survey. The archaeologist wasbriefly instructed to take the photos from the archaeolo-gist’s point of view, but having some concern regardingthe overlapping of features. After analysing the pho-tos, some of them were excluded, because they con-tained problematic materials, such as glass and shinystainless steel from the site protection structure, whichcould compromise the accuracy of the reconstruction.Therefore, for reconstruction purpose 34 photos werevalidated. The elapsed time for this initial step was ap-proximately 30 minutes.



Figure 8: Image-based 3D reconstruction of the Fonte do Ídolo:Surface mode (left) and Wireframe mode (right)

The first 3D reconstruction was executed with a loopof 17 photos which were directly uploaded using the123DCatch desktop application and it took about 20minutes to produce a textured 3D mesh of the Fontedo Ídolo. The obtained mesh was composed of 97,362vertices and 193,037 triangles. A second 17 photos loopwas added to the first set of photos to improve the initialmesh quality and it took about 30 minutes to achieve asignificant enhancement of the mesh: 137,992 verticesand 266,883 triangles. Since there were no more photosto use, at this point the refinement of the mesh had to becarried out by stitching manually a set of vertices. Fivephotos with noteworthy features were carefully selectedto stitch manually 3D vertices. A set of 9 vertices weredetermined and afterwords the scene was again submit-ted for processing. The final mesh was improved with162,351 vertices and 323,131 triangles (see figure 8).The entire workflow took less than a day.

4 EVALUATION OF RESULTSThe number of 3D vertices and consequently the meshdensity of Milreu’s Roman pool 3D reconstructionwere much higher when using VisualSfM combinedwith MeshLab. However, generally it takes more timeto obtain the final reconstructed object, since it dependsgreatly on the computer hardware. Also, regardingthe graphical user interface, 123DCatch is more userfriendly. In fact, besides the lower mesh density, themain disadvantages of 123DCatch are the constantneed of an internet connection (not always possible inan archaeological site) and the limited possibility toparametrize the application regarding the mesh densityof the chosen archaeological object or structure. Bothsoftware packages are not able to perform surfacesegmentation in a simple and prompt way, whichis necessary to define the stratigraphic units of thearchaeological site.

In the case of the Fonte do Ídolo, there is a clear ad-vantage of the image-based survey over the traditional

archaeological survey regarding the human resourcesthat were used and the elapsed time. Also, the visualquality and the three-dimensionality of the image-basedreconstruction strongly enhance the perception of thesite: the different elements and structures are naturallyidentified even by non-experts. Moreover, there is anacceptable error regarding the measures between struc-tures and the dimensions of the site elements. Also, thedataset obtained with the traditional survey is not well-suited for 3D visualization purposes due to its 2D na-ture. However, the 2D representation of archaeologicaldata is the one the archaeologists are more familiarizedwith.

No doubt that the laser survey, even after its simplifica-tion with NSA, retains a better visual quality than theimage-based survey regarding the façade of the Fontedo Ídolo. Of course, considering the overall 3D recon-struction, the image-based survey preserves a more bal-anced visual accuracy, partially achieved due to the tex-tures generated during 123DCatch’s 3D reconstruction.Also, the laser survey needs skilled labor and more hu-man resources than the image-based data acquisition.The time required for both surveys to obtain a 3D rep-resentation of the Fonte do Ídolo is substantially equiv-alent. The laser scanner used within this survey doesnot seem to be the most adequate for large site surveys;however it is quite satisfactory to be used in confinedregions where more detail is required, such as the sculp-tural and epigraphical elements of the façade.

Both non-traditional surveys produce 3D datasets thatare adequate for 3D representation and visualization,used during the interpretation and research stages of thearchaeological process. There are several adequate vi-sualization systems that can be used, however, for theseexamples the open source Visualization Toolkit (VTK)from Kitware Inc. is used. This toolkit was preferredsince it supports several visualization and modellingtechniques that are useful for interpretation purposes inarchaeology. In fact, using VTK it is possible to rep-



Figure 9: Arbitrary section of the Fonte do Ídolo

resent the archaeological site as a virtual scale modelwhich can be freely manipulated. Figure 9 shows a 3Darbitrary plane that freely defines a section over the 3Drepresentation of the Fonte do Ídolo.

Image-based 3D models are also acceptable for dis-semination purposes, which is one of the last stages ofthe archaeological process. The 3D representation ofthe Fonte do Ídolo was processed with the Instant Re-ality framework developed at Fraunhofer IGD–VCST[Inst14], which offers various components that, on theone hand, simplify the geometry of a 3D model and, onthe other hand, convert the input file into a X3D file. Inthis way, it is straightforward to use the X3DOM frame-work to make the model available on the Web (see fig-ure 10) and to manipulate the 3D content. A clear ad-vantage of using this framework is that web-browserswhich support X3DOM do not require to install a plu-gin to visualize 3D models.

Figure 10: X3D file of the Fonte do Ídolo

If the aim is only to visualize the 3D model on the Web,another dissemination possibility is to publish the 3Dreconstruction on Sketchfab [SFab14]. Sketchfab is afree instant in-browser viewer that does not require a

plugin and which can be embedded on a web-page (seefigure 11).

Figure 11: The Fonte do Ídolo integrated on Sketchfab

5 CONCLUDING REMARKSRegarding the 3D reconstruction of the quadrangularpool at the Roman Villa in Milreu using two differentimage-based modelling packages, it is clear that bothare valid options. However, in spite of the higher meshdensity of the VisualSfM and MeshLab reconstruction,it is preferable to use 123DCatch because of the (1)user-friendlier graphical interface, the (2) ellapsed re-construction time and the (3) very acceptable visualquality.

The distinct methodologies for archaeological surveycarried out at the Fonte do Idolo illustrate that usingimage-based modelling for surveying an archaeologi-cal site is adequate. It generates a 3D model with thenecessary visual accuracy and quality for the archaeo-logical record and for interpretation and disseminationpurposes. However, this does not mean (at least fornow) that the traditional archaeological survey can becompletely substituted by an image-based archaeolog-ical survey. In fact, for sites where it is necessary to



record shiny materials, such as glass or metal, the 3Dreconstruction might experience some difficulties.

The segmentation of an image-based reconstructionstill remains an unwanted constraint for archaeology.In fact, a major purpose for future work is to developa segmentation tool, based on VTK, to define thestratigraphic units that are indispensable for archaeo-logical analysis and interpretation, and a volumetricreconstruction module to represent the volumes definedby the stratigraphic units.

6 REFERENCES[123D] Autodesk. (2013, August) Autodesk 123D -

123D catch turn photos into 3D models. [On-line]. Available: http://www.123dapp.com/catch

[Arn07a] D. Arnold and G. Geser, Research Agendafor the Applications of ICT to Cultural Heritage.EPOCH, 2007.

[Bar08a] J. Barbosa, P. Bernardes, M. Almeida,P. Gomes, R. Goncalves, and A. Proenca, “A tech-nology cocktail for a 3D photo-realistic model ofa I century roman fountain: Range scanning, RTIand physically based rendering,” in Proceedingsof VAST’08. Eurographics Association, 2008,pp. 25–32.

[Bru12a] M. L. Brutto and P. Meli, “Computer visiontools for 3D modelling in archaeology,” Interna-tional Journal of Heritage in the Digital Era, vol.1, pp. 1–6, 2012.

[Car08a] J. Carrasco, A. Ugalde, S. Dils, Á. Ro-dríguez, J. Oliveira, J. F. de Rueda Roigé, andN. Vidal Teruel, A rota do mosaico romano. OSul da Hispânia (Andalzia e Algarve): Cidadese villae notáveis de Bética e Lusitânia romanas,Departamento de História, Arqueologia e Patri-monio da Universidade do Algarve, 2008.

[Cig08a] P. Cignoni, M. Callieri, M. Corsini,M. Dellepiane, F. Ganovelli, and G. Ranzuglia,“Meshlab: an open-source mesh processing tool,”in Eurographics Italian Chapter Conference. Eu-rographics Association, 2008, pp. 129–136.

[DeR13a] J. D. Reu, G. Plets, G. Verhoeven, P. D.Smedt, M. Bats, B. Cherretté, W. D. Maeyer,J. Deconynck, D. Herremans, P. Laloo, M. V.Meirvenne, and W. D. Clercq, “Towards a three-dimensional cost-effective registration of the ar-chaeological heritage,” Journal of ArchaeologicalScience, vol. 40, no. 2, pp. 1108–1121, 2013.

[DeR14a] J. D. Reu, P. D. Smedt, D. Herremans, M. V.Meirvenne, P. Laloo, and W. D. Clercq, “On intro-ducing an image-based 3D reconstruction methodin archaeological excavation practice,” Journal

of Archaeological Science, vol. 41, pp. 251–262,2014.

[Ele08a] A. G. Elena, R. Mar, and M. Martins, A Fontedo Ídolo: análise, interpretação e reconstituiçãodo santuário, ser. Bracara Augusta. Unidade deArqueologia da Universidade do Minho, 2008.

[Fru07a] F. Silva, “NSA simplification algorithm: geo-metrical vs. visual quality,” in International Con-ference on Computational Science and its Ap-plications, 2007. ICCSA 2007. , Aug 2007, pp.515–523.

[Hau02a] T. Hauschild and F. Teichner, Milreu: ruí-nas, ser. Roteiros da Arqueologia Portuguesa. In-stituto Português do Património Arquitectónico,2002.

[Her12a] S. Hermon, H. Khalaily, I. Milevski, N. Am-ico, G. Iannone, and N. Getzov, “Archaeologicalfield documentation and architectonic analysis - a3D approach. Ein Zippori as study case,” in Pro-ceedings of VAST2012, D. B. Arnold, J. Kaminski,F. Niccolucci, and A. Stork, Eds. EurographicsAssociation, 2012, pp. 113–120.

[Inst14] Fraunhofer IGD. (2014, March) InstantRe-ality. [Online]. Available: http://www.instantreality.org/

[McC14a] J. McCarthy, “Multi-image photogramme-try as a practical tool for cultural heritage surveyand community engagement,” Journal of Archae-ological Science, vol. 43, pp. 175–185, 2014.

[Nic10a] F. Niccolucci, “Technologies, standards andbusiness models for the formation of virtual col-lections of 3d replicas of museum objects: The3D-Coform project,” in IST-Africa, 2010, May2010, pp. 1–8.

[Pol01a] M. Pollefeys, L. Van Gool, M. Vergauwen,K. Cornelis, F. Verbiest, and J. Tops, “Image-based 3D acquisition of archaeological heritageand applications,” in Proceedings of VAST’01.ACM, 2001, pp. 255–262.

[San13a] C. Santagati, L. Inzerillo, and F. D. Paola,“Image-based modeling techniques for architec-tural heritage 3D digitalization: Limits and po-tentialities,” International Archives of the Pho-togrammetry, Remote Sensing and Spatial Infor-mation Sciences, vol. XL-5, no. w2, pp. 555–560,September 2013.

[SFab14] Sketchfab. (2014, March) Sketchfab - Theweb platform for publishing your interactive3D models. [Online]. Available: https://sketchfab.com/

[Wu13] C. Wu. (2013, August) VisualSfM : A visualstructure from motion system. [Online]. Avail-able: http://homes.cs.washington.edu/~ccwu/vsfm/



WSCG 2014 - repositorium.sdum.uminho.pt

Documents

Transcript of WSCG 2014 - repositorium.sdum.uminho.pt