Transforming Intangible Folkloric Performing Arts into Tangible

Choreographic Digital Objects: The Terpsichore Approach

Anastasios Doulamis, Athanasios Voulodimos, Nikolaos Doulamis, Sofia Soile

and Anastasios Lampropoulos

Photogrammetry and Computer Vision Lab, National Technical University of Athens,

Zografou Campus, 15780, Athens, Greece

Keywords: Intangible Cultural Heritage, Performing Arts, Folkloric Dances, 3D Modelling, Cultural Heritage

Digitisation.

Abstract: Intangible Cultural Heritage is a mainspring of cultural diversity and as such it should be a focal point in

cultural heritage preservation and safeguarding endeavours. Nevertheless, although significant progress has

been made in digitization technology as regards tangible cultural assets and especially in the area of 3D

reconstruction, the e-documentation of intangible cultural heritage has not seen comparable progress. One of

the main reasons associated lies in the significant challenges involved in the systematic e-digitisation of

intangible cultural assets, such as performing arts. In this paper, we present at a high-level an approach for

transforming intangible cultural assets, namely folk dances, into tangible choreographic digital objects. The

approach is being implemented in the context of the H2020 European project “Terpsichore”.

1 INTRODUCTION

Intangible Cultural Heritage (ICH) content means

“the practices, representations, expressions,

knowledge, skills – as well as the instruments,

objects, artefacts and cultural spaces associated

therewith”. Intangible Cultural Heritage (ICH) is a

very important mainspring of cultural diversity and a

guarantee of sustainable development, as underscored

in the UNESCO Recommendation on the

safeguarding of Traditional Culture and folklore of

1989, in the UNESCO Universal Declaration on

Cultural Diversity of 2001 and in the Convention for

the Safeguarding of the Intangible Cultural Heritage

(Kyriakaki, 2014). Improving the digitization

technology regarding capturing, modelling and

mathematical representation of performance arts and

especially folklore dances is critical in: (i) promoting

cultural diversity to the children and the youth

through the safeguard of traditional performance arts;

(ii) making local communities and especially

indigenous people aware of the richness of their

intangible heritage; (iii) strengthening cooperation

and intercultural dialogue between people,

different cultures and countries.

Although ICH content, especially traditional

folklore performing arts, is commonly deemed

worthy of preservation by UNESCO and the EU

Treaty, most of the current research efforts focus on

tangible cultural assets, while the ICH content has

been given less emphasis. The primary difficulty

stems by the complex structure of ICH, its dynamic

nature, the interaction among the objects and the

environment, as well as emotional elements (e.g., the

way of expression and dancers’ style). Of course there

have been some notable efforts such as the i-

Treasures project which provides a platform to access

ICH resources and contribute to the transmission of

rare know-how from Living Human Treasures to

apprentices (Dimitropoulos, 2016) and the RePlay

project, whose goal is to understand, preserve, protect

and promote traditional sports (Linaza, 2013).

Towards this direction, the Terpsichore project

aims to study, analyse, design, research, train,

implement and validate an innovative framework for

affordable digitization, modelling, archiving, e-

preservation and presentation of ICH content related

to folk dances, in a wide range of users (dance

professionals, dance teachers, creative industries and

general public).

Exploring the digitization technology regarding

folklore performances constitutes a significant impact

Doulamis A., Voulodimos A., Doulamis N., Soile S. and Lampropoulos A.

Transforming Intangible Folkloric Performing Arts into Tangible Choreographic Digital Objects: The Terpsichore Approach.

DOI: 10.5220/0006347304510460

at European level. On one hand, the multi-cultural

intangible dimension of Europe is documented,

preserved, made available to everybody on Internet.

On the other hand, the multifaceted value to the ICH

content for usage in education, tourism, art, media,

science and leisure settings is added.

Currently, digital technology has been widely

adopted, which greatly accelerates efforts and

efficiency of Cultural Heritage (CH) preservation and

protection. At the same time, it enhances CH in the

digital era, creating enriched virtual surrogates. Many

research works have been proposed in the literature

on archiving tangible cultural assets in the form of

digital content (Li, 2010). Although the

aforementioned significant achievements for

improving the digitization technology towards a more

cost-effective automated and semantically enriched

representation, protection, presentation and re-use of

the CH via the European Digital Library

EUROPEANA, very few efforts exist in creating

breakthrough digitization technology, improving the

e-documentation (3D modelling enriched with

multimedia metadata and ontologies), the e-

preservation and re-use of ICH traditional music and

fashion, folklore, handicraft, etc.

Terpsichore targets at integrating the latest

innovative results of photogrammetry, computer

vision, semantic technologies, time evolved

modeling, combined with the story telling and

folklore choreography. An important output of the

project will be a Web based cultural server/viewer

with the purpose to allow user’s interaction,

visualization, interface with existing cultural libraries

and enrichment functionalities to result in virtual

surrogates and media application scenarios that

release the potential economic impact of ICH. The

final product will support a set of services such as

virtual/augmented reality, social media, interactive

maps, presentation and learning of European Folk

dances with significant impact on the European

society, culture and tourism.

The remainder of this paper is structured as

follows: In Section 2 we review the state of the art in

the fields pertaining to the Terpsichore approach,

which is presented in Section 3. Finally, Section 4

concludes the paper.

2 RELATED WORK

Although significant progress has been made in

digitization technology as regards tangible cultural

assets and especially in the area of 3D reconstruction

– see the research achievements of the projects 3D-

COFORM (www.3d-coform.eu), EPOCH

(www.epoch-net.org), IMPACT (www.impact-

project.eu), PRESTOSPACE (prestospace.org) – the

e-documentation of intangible cultural assets is not

yet evident, especially in the case of folklore

performing arts. This is mainly due to the complex

multi-disciplinarity of the folklore performances

which presents a series of challenges ranging from the

choreography, the folk music, the –uniforms, -music

and from the digitization and computer vision to

spatio-temporal (4D) dynamic modelling and virtual

scene generation as discussed above. It is important

to mention that this is the first time in the entire

European Union that a Knowledge Alliance of this

calibre is trying to undertake such innovative project,

which aims to act as a pioneering mechanism for

unifying ICH content with already existing digitized

CH content from digital libraries (such as the folklore

stories documented in the different e-records in

EUROPEANA), and whose outcomes will not only

lead to advanced scientific publications, but also

patents, which will boost EU economic growth. In the

following, we present the current e-documentation

technologies and the respective limitations that the

Terpsichore project aims to address.

2.1 3D Data Acquisition and Processing

The most popular methods for 3D capturing are

divided into two main categories: active methods

(laser scanners, range finders, structured light

projectors) and passive methods (stereo vision and

visual hulls). The most common used passive method

is to attach distinctive markers to the body of a human

and track these markers in images acquired by

multiple calibrated cameras (www.vicon.com). In

this case, method's accuracy is limited by the number

of markers available. Markerless capture methods

(Carceroni, 2001; Li, 2008; Pons, 2006) based on

computer vision technology overcome this problem.

However, these approaches do not fully exploit global

spatiotemporal consistency constraints and are

susceptible to error approximation.

The method of (Furukawa, 2007) addresses these

limitations, however, it relies on PVMS software

efficiency. In (Yamasaki, 2010) a system using 22

cameras is proposed. 3D modeling is based on the

combination of volume intersection and stereo

matching. However, the original shape distribution

cannot be generated stably. Active methods offer

higher stability and accuracy compared to passive

methods. In (Cui, 2010) a 3D shape scanning system

based on the ToF camera is presented. Although,

these cameras are of low cost, they present

limitations, such as very low X/Y resolution and

random noise behavior. In (Sakashita, 2011) a system

for capturing textured 3D shapes is presented, by

using a multi-band camera in combination with an

infrared structured light projector. However, it

requires no other illumination in the environment.

FusionKinect (Izadi, 2011) uses a Kinect camera to

generate real-time depth maps containing discrete

range measurements of the physical scene. However,

depth data are inherently noisy and depth maps

contain “holes” where no readings were obtained.

After capturing, computer vision techniques are

necessary for data reduction, filtering, optical flow

and disparity estimation. The techniques used to solve

correspondence problems are similar and can be

categorized as energy-based and feature-based.

Energy-based methods (Alvarez, 2002) minimize a

cost function plus a regularization term, in the

framework of (Horn, 1981), to solve for the 2D

displacements and yield very accurate, dense flow

fields. However, they fail as displacements get too

large. Feature-based methods (Pons, 2006; Furukawa,

2011; Shrivastava, 2011; Liu, 2011) match features in

different images. This kind of methods are able to

overcome the problem of large displacements by

using the concept of coarse-to-fine image warping,

however, this downsampling removes information

that may be vital for establishing correct matches.

Data processing techniques will be used to refine

acquired data (remove noises, remove “holes”,

accelerate 3D registration). The method of (Mitra,

2004) based on local least square fitting for

estimating the normals at all sample points of a point

cloud data set, in the presence of noise. The work of

(Ruhnke, 2012) proposes an approach to obtain

highly accurate 3D models from range data by jointly

optimize the poses of the sensor and the positions of

the surface points measured with a range scanning

device. The work of (Rusu, 2009) uses Point Feature

Histograms for accelerating 3D registration problem.

2.2 3D Modelling and Rendering

Modelling is a process to create a model, which by

definition is an abstract representation that reflects the

characteristics of a given entity either physical or

conceptual. 3D modelling relies on computational

geometry techniques such as skeleton extraction,

division of space into subspaces and mesh

reconstruction. The authors of (Menier, 2006) present

an approach to recover body motions from multiple

views using a 3D skeletal model, which is an a priori

articulated model consists in kinematic chain of

segments representing a body pose. In (Chen, 2009)

an approach for simultaneously reconstructing 3D

human motion and full-body skeletal size from a

small set of 2D image features is presented. It

resolves the ambiguity for skeleton reconstruction

using pre-recorded human skeleton data. The

approach of (Gall, 2009) recovers the movement of

the skeleton, as well as, the possibly non-rigid

temporal deformation of the 3D surface by using an

articulated template model and silhouettes from a

multi-view image sequence. Another volumetric

approach is this of (Matsuyama, 2004) that uses

silhouette volume intersection to generate the 3D

voxel representation of the object shape and uses a

high fidelity texture mapping algorithm to convert the

3D object shape into a triangular patch representation.

3D rendering is a necessary process for

visualizing modelled content. However, real-time

rendering of detailed animated characters, especially

in crowded simulations like dance, is a challenging

problem in computer graphics. Textured polygonal

meshes provide high-quality representation at the

expense of a high rendering cost. To overcome this

problem, several techniques focusing on providing

level-of-detail representations have been proposed.

Image-based pre-computed impostors (Tecchia,

2002) render distant characters as a textured polygon

to accelerate rendering of animated characters. A

much more memory-efficient but view-dependent

approach is to subdivide each animated character into

a collection of pieces, in order to use separate

impostors for different body parts (Kavan, 2008). In

(Pettré, 2006) a three-level-of-detail approach is

described, combining the animation quality of

dynamic meshes with the high performance offered

by static meshes and impostors. The technique in

(Andújar, 2007) adopts a relief mapping approach to

encode details in arbitrary 3D models with minimal

supporting geometry.

2.3 Symbolic Representation,

Ontologies and Harvesting

During dancing performances, motion gestures are

used to communicate a storyline in an aesthetically

pleasing manner. Although, humans automatically

perceive and understand such gestures, from the point

of view of computer science these gestures have to be

analyzed under an appropriate framework with

appropriate features, such as repetitive patterns,

motion trajectories and motion inclusions, in order to

extract their semantics. In the work of (Moon, 2008)

a generative statistical approach to human motion

modeling and tracking that utilizes probabilistic latent

semantic analysis to describe the mapping of image

features to 3D human pose estimate, is presented. The

latent variables describe intrinsic motion semantics

linking human figure to 3D pose estimates. In (Yang,

2010) the dance motion is analyzed to extract the

repetitive patterns and compute prerequisite relations

among them. These relations are illustrated by a

concept map that is constructed automatically. The

authors of (Kahol, 2004) based on human anatomy

propose a method for deriving choreographer

segmentation profiles of dance motion capture

sequences. A more stable and descriptive framework

for human motion analysis is the Laban Movement

Analysis (LMA) framework. LMA has been proposed

and used from the viewpoint of the analysis of body

motions and it can be used for not only motion

analysis but also extracting and generating expression

of movements in general (Aristidou, 2014a;

Aristidou, 2014b). The method of (Bouchard, 2007)

uses LMA Effort component as a basis for motion

capture segmentation, which is more meaningful than

kinematic features, and it is easier to compute for

general motions than semantic features. In (Woo,

2000) LMA has been used to obtain dancer's intention

and estimate emotional or sensitivity information of

the dance performance. In (Santos, 2011) the

performance of different signal features regarding the

qualitative meaning of LMA semantics is examined.

Especially, this work is based on the study of body

part trajectories and the objective is to apply multiple

feature generation algorithms to segment the signals

according to LMA theory, in order to find patterns

and define the most prominent features in each of the

descriptors defined on Labanotation (Chi, 2000).

“Synchronous Objects” project

(synchronousobjects.osu.edu) is focusing on the topic

of transforming choreographies into symbolic

representations. However, the dancers have been

interactively marked by animators frame by frame

with the aim to track their movements. Finally, within

the project “MotionBank” (motionbank.org) the

synchronous object is followed up with the aim to

create “online scores” visualizing and illustrating the

choreograph’s intention.

3 4D MODELLING OF ICH: THE

TERPSICHORE APPROACH

To achieve a reliable 4D modelling of intangible

cultural heritage (ICH) assets, such as dances, a new

pioneer framework should be adopted. The research

tools needed to be applied are depicted in Figure 1.

The goals are to: (a) explore a scalable capturing

framework as regards 3D reconstruction of ICH in

terms of accuracy and cost-effectiveness, (b) develop

the captured visual 3D signals into symbolic data that

represent the overall human creativity, (c) define an

interoperable Intangible Cultural Metadata Interface

(ICMI) for folklore performing arts, and (d) nominate

the appropriate codification of the extracted symbolic

data structures in an interoperable form that permits

interconnection with existing specifications of

national digital cultural repositories or international

like the EU digital library EUROPEANA

(www.europeana.eu/portal/) and the UNESCO

Memory of the World (Figure 2).

During the 20th century there have been several

attempts to model human creativity in performing

arts. In 1920s Rudolf Laban developed a system of

movement notation that eventually evolved into

modern-day Laban Movement Analysis (LMA)

(Pforsich, 1977)0, which provides a language for

describing, visualizing, interpreting, and

documenting all varieties of human movement, in an

attempt to preserve classic choreographies. In the

early 2000s LMA extensively used for analyzing

dance performances (Davis, 2001) and create digital

archives of dancing in the area of education and

research.

The main goal of the Terpsichore project is to

leverage such past efforts towards the transformation

of intangible cultural heritage content to 3D virtual

content, through the development and exploitation of

affordable digitization technology. Towards this

direction, fusion of different scientific and

technological fields, such as capturing technology,

computer vision and learning, 3D modeling and

reconstruction, virtual reality, computer graphics and

data aggregation for metadata extraction, is

necessary. There are four main distinct research

directions in the context of the project, which will be

presented in the following.

3.1 Choreographic Analysis, Design

and Modelling

The main research objective in this section is to

analyze the spatial specifications, the attributes and

the properties of folklore traditional performing arts.

The analysis describes all the aspects that are

needed for recording human creativity based on

tradition; thus apart from the human movement, the

way of expression, the associated emotional

characteristics, the style as well as additional

contextual data, such as climate conditions, social-

cultural factors, stylistic variations, the accompanied

untold scenarios and stories should be recorded. All

Figure 1: The main research components towards an intangible cultural heritage content (ICH) digitalization.

Figure 2: Framework of the proposed intangible cultural content digitalization.

Figure 3: Description of the basic metadata elements used for representing the folklore performing arts.

dance

uniforms

music scores & lyrics

Folklore

3D TV

The Proposed

Scheme

Advisory

Board

Metadata Elements

Instantiation

Metadata Structure

Europeana

UNESCO

Metadata

Specifications

Alignment

the aforementioned specifications will be surveyed in

such a way to derive an interoperable description

framework based on which we are able to design and

define the Intangible Cultural Metadata Interface

(ICMI). ICMI aims at specifying a set of metadata

that are necessary for representing the rich intangible

cultural heritage information and especially in case of

folklore traditional performing arts. In addition, ICMI

introduces an interoperable description scheme

framework able to specify the metadata structure and

the relations between the extracted metadata. Thus,

ICMI specifies not only the appropriate metadata for

representing human creativity but also the structure

and semantics of relations between its components.

Figure 3 presents an indicative description framework

of the basic metadata elements used for the

description of folklore performing arts. As is

observed, the metadata of the ICMI are discriminated

into four main categories; the low level feature

metadata, the contextual and environmental

metadata, the socio-cultural factors as well as the

emotional attributes.

The pool of the basic metadata used for describing

the intangible cultural assets of traditional folklore

performing arts are framed with a metadata structure

format able to interoperably represent the semantics

relations between the metadata components of Figure

3. Special emphasis should be given in order to align

the ICMI format with existing specifications of digital

cultural libraries, such as EUROPEANA and

UNESCO the Memory of World, since this will allow

the easy archiving, usage and re-usage of the digitized

intangible cultural content with the content of large

digital cultural repositories.

3.2 Capturing and 3D Modelling of

Static Objects

Another important research issue deals with

technologies able to capture and virtually 3D

reconstruct traditional performing arts under an

affordable, cost-effective and accurate digitization

framework. To do this, an innovative technological

framework able to combine advanced technologies in

the area of photogrammetry and computer vision

should be introduced. A scalable capturing

framework that combines 3D modelling technology

coming by a set of multi-view stereo camera

architecture and a low cost depth sensors (Kinect) is

necessary to be adopted. The balance of using multi-

view stereo imaging (high resolution cameras plus

dense image matching techniques) and low cost depth

sensors able to generate depth information in real

time are obtained in terms of accuracy and cost-

effectiveness. In other words, the fusion of the

information from different sensors in order to

increase the accuracy or reduce the number of

cameras necessary for data capturing remains a

challenge.

As regards the multi-view imaging architecture,

the 3D information is extracted by applying dense

image matching techniques. For each stereo model

observing common content a correspondence is

determined for each pixel individually. By using

these correspondences between all stereo models, all

3D Points can be triangulated based on their viewing

rays at once. This leads to a very dense and accurate

point cloud. In order to acquire the movements in

time, this step is performed for each frame in time for

all synchronized cameras – leading to 3D point clouds

for each time stamp. However, this is a computational

intensive process that increases the total cost of

digitization. 3D modelling tools appropriately

designed for time varying shapes should be used.

Such methodologies exploit motion information as

well as tracking methods Subsequently, a volumetric

integration of this depth information not only enables

the extraction of a volumetric representation, but also

to fill small gaps and reduce noise.

The fusion process should be performed under a

calibrated network of cameras or depth sensors.

Network calibration is very important since it allows

the implementation of super-resolution methods and

detection of confidence data as obtained either by the

depth sensors or the multi-view imaging. The use of

self-calibration methodologies able to automatically

calibrate the network of cameras and depth sensors

using computer vision tools is a major research issue.

Figure 4 presents the aforementioned methodology.

3.3 3D Modelling of Moving Objects

Another research field includes imaging

methodologies based on the combination of

photogrammetric, computer vision and computer

graphics techniques able to automate the 3D

modelling procedure of moving objects. Towards this

direction, initially, segmentation algorithms able to

isolate the foreground objects from the background is

applied. This step is critical since it allows the human

objects to be separated from the background content.

The foreground objects are dynamically updated

through the motion estimation captured by the

capturing layer. On the contrary, background content

is updated by the information provided by the depth

sensor network. A set of deformable models is used

to describe the human movement. These set of

deformable models are provided dealing with the

Figure 4: A representation of the methodology resulting in a scalable capturing framework in terms of accuracy and

performance.

Figure 5: The 3D modelling methodology for moving objects.

analysis, design and modelling of folklore traditional

performing arts.

In order to allow a cost effective 3D digitization

process, a voxel 3D skeletonization is performed. In

this way, we are able to reduce the amount of

information needed to describe the human object;

therefore, a reduction of the cost of processing

resulting in more accurate and cost-effective 3D

modelling of moving objects is possible. 3D skeletons

are medial axes transform and encode mostly motion

information. 3D skeletons separate motion estimation

from shape and thus, they allow a more accurate 3D

model updating through time.

The derived 3D skeletons are tracked in time

using motion capture and tracking computer vision

methodologies. Tracking process is assist through the

selection of appropriate geometrically enriched data.

By taking into consideration motion tracking, which

is performed on the 3D skeletons to increase accuracy

and computational efficiency, as well as the set of

deformable models appropriate for a specific folklore

performance, we are able to automatically update the

Visual Sensor

Multi-view

camera design

Depth sensor

network design

Network

calibration

Geometrically-

based visual

analysis

Image Matching

Confident data

Creation of

super-resolution

Data fusion

Volumetric

Surface

Intergation

The Scalable Capturing Framework

Static 3D

models

Foreground/

background

separation

3D Voxel

Skeletonization

Motion Capture

and Tracking

Deformable

models &

Surface

updating

Model Refinement

(ToF information)

ToF information

Statio-temporal

positioning

Deformable

models

Visual features

Spatio-temporal positioning

detected 3D models to fit the properties of the current

human movement and shape constraints as obtained

by the static 3D models.

Despite the efficiency of the aforementioned

methodology, possible errors (in the 3D

skeletonization and tracking process) generate

erroneous virtual 3D reconstructions. To address this

difficult, we enhance the results of the 3D modelling

of moving objects using information derived from the

depth sensor network. In this case, we are able to

exploit the depth information to improve 3D

modelling accuracy. Figure 5 depicts this

architecture.

3.4 Symbolic Representation and

Extraction of Semantic Signatures

3D modelling is critical for encoding the complex 3D

reconstructions of performing arts into a set of

compact semantic signatures in a similar way that a

music song is encoded using a music score. For this

reason, computational geometry algorithms are used

to decode the spatial-temporal trajectory of the

performances. This includes methodologies for

positioning both in 3D and temporal space. Then,

semantic signatures and respective spatial-temporal

associations are extracted to represent the

performances with high level concepts. It is clear that

semantic analysis aids the digitization and computer

vision process and vice versa.

The visualization of scores fosters the

understanding of the dance and it helps visitors,

dancers and choreographs to comprehend the

structure and the intension of the dance. A more

formalized documentation of the dance will be

supported with an automated mapping of capturing

data to formal and abstract chorographic notations

(“Symbolic Representations”). Hereby, the captured

dance could be coded using traditional approaches

like the Laban Motion Analysis or modern ballet

notation forms like the “peacemaker” from William

Forsythe and David Kern.

3.5 Virtual Scene Generation

Finally, the complete reconstructed 4D-scenario can

be visualized within an interactive Web-Viewer.

Virtual scene generation exploits both the produced

3D models, but also the codified symbolic

representations, generating an innovative and unique

research framework able to allow for manipulation,

usage and re-usage of the cultural objects. This

interactive Web-Viewer links the extracted Symbolic

Representations to the 4D-reconstruction of the

choreography. Within this Web-Viewer beside

different viewpoints also the perspective of a specific

dance can be chosen.

4 CONCLUSIONS

In this paper we have presented the concept of the

Terpsichore project. The research directions of the

project will span across various fields, including

choreographic analysis, 3D data capturing, 3D

modelling of static and moving objects and symbolic

representations, among others. Through the described

approach Terpsichore aims to study, analyse, design,

research, train, implement and validate an innovative

framework for affordable digitization, modelling,

archiving, e-preservation and presentation of

Intangible Cultural Heritage content related to folk

dances, in a wide range of users, including dance

professionals, dance teachers, creative industries and

the general public. Future

ACKNOWLEDGEMENTS

This work has been supported by the H2020-MSCA-

RISE project “Transforming Intangible Folkloric

Performing Arts into Tangible Choreographic Digital

Objects (Terpsichore)” funded by the European

Commission under grant agreement no 691218. The

authors would like to help all partners for their

contribution and collaboration.

REFERENCES

Alvarez, L., R. Deriche, J. Sánchez, and J. Weickert, 2002.

“Dense Disparity Map Estimation Respecting Image

Discontinuities: A PDE and Scale-Space Based

Approach,” Journal of Visual Communication and

Image Representation, vol. 13, no. 1–2, pp. 3–21, 2002.

Andújar, C., J. Boo, P. Brunet, M. Fairén, I. Navazo, P.

Vázquez, and À. Vinacua, 2007. “Omni-directional

Relief Impostors,” Computer Graphics Forum, vol. 26,

no. 3, pp. 553–560, 2007.

Aristidou A. & Y. Chrysanthou, “Feature extraction for

human motion indexing of acted dance performances.”

In Proceedings of the 9th International Conference on

Computer Graphics Theory and Applications, 2014 (pp.

277-287).

Aristidou, A., E. Stavrakis, & Y. Chrysanthou, “LMA-

Based Motion Retrieval for Folk Dance Cultural

Heritage.” In Digital Heritage. Progress in Cultural

Heritage: Documentation, Preservation, and Protection,

2014, (pp. 207-216). Springer International Pub

Bouchard D. and N. Badler, “Semantic Segmentation of

Motion Capture Using Laban Movement Analysis,” in

Proceedings of the 7th international conference on

Intelligent Virtual Agents, Berlin, Heidelberg, 2007,

pp. 37–44.

Carceroni, R.L. and K. N. Kutalakos, 2001. “Multi-view

scene capture by surfel sampling: from video streams to

non-rigid 3D motion, shape and reflectance,” in Eighth

IEEE International Conference on Computer Vision,

2001. ICCV 2001. Proceedings, 2001, vol. 2, pp. 60 –

67 vol.2.

Chen, Y.-L. and J. Chai, 2010. “3D Reconstruction of

Human Motion and Skeleton from Uncalibrated

Monocular Video,” in Computer Vision – ACCV 2009,

H. Zha, R. Taniguchi, and S. Maybank, Eds. Springer

Berlin Heidelberg, pp. 71–82.

Chi, D. M. Costa, L. Zhao, and N. Badler, “The EMOTE

model for effort and shape,” in Proceedings of the 27th

annual conference on Computer graphics and

interactive techniques, New York, NY, USA, 2000, pp.

173–182.

Cui, Y., S. Schuon, D. Chan, S. Thrun, and C. Theobalt,

2010. “3D shape scanning with a time-of-flight

camera,” in 2010 IEEE Conference on Computer

Vision and Pattern Recognition (CVPR), 2010, pp.

1173 –1180.

Davis, E., 2001. “BEYOND DANCE Laban's Legacy of

Movement Analysis,” Brechin Books, 2001

Dimitropoulos, K., P. Barmpoutis, A. Kitsikidis, and N.

Grammalidis, 2016. "Extracting dynamics from multi-

dimensional time-evolving data using a bag of higher-

order Linear Dynamical Systems", in Proc. 11th

International Conference on Computer Vision Theory

and Applications (VISAPP 2016), Rome, Italy.

Dobbyn, S., J. Hamill, K. O’Conor, and C. O’Sullivan,

2005. “Geopostors: a real-time geometry / impostor

crowd rendering system,” in Proceedings of the 2005

symposium on Interactive 3D graphics and games, New

York, NY, USA, 2005, pp. 95–102.

Furukawa, Y. and J. Ponce, 2010. “Dense 3D Motion

Capture from Synchronized Video Streams,” in Image

and Geometry Processing for 3-D Cinematography, R.

Ronfard and G. Taubin, Eds. Springer Berlin

Heidelberg, 2010, pp. 193–211.

Gall, J, C. Stoll, E. de Aguiar, C. Theobalt, B. Rosenhahn,

and H.-P. Seidel, 2009. “Motion capture using joint

skeleton tracking and surface estimation,” in IEEE

Conference on Computer Vision and Pattern

Recognition, CVPR 2009, 2009, pp. 1746 –1753.

Horn B. K. P. and B. G. Schunck, 1981. “Determining

optical flow,” Artificial Intelligence, vol. 17, no. 1–3,

pp. 185–203.

Izadi, S., D. Kim, O. Hilliges, D. Molyneaux, R.

Newcombe, P. Kohli, J. Shotton, S. Hodges, D.

Freeman, A. Davison, and A. Fitzgibbon, 2011.

“KinectFusion: real-time 3D reconstruction and

interaction using a moving depth camera,” in

Proceedings of the 24th annual ACM symposium on

User interface software and technology, New York,

NY, USA, 2011, pp. 559–568.

Kahol, K., P. Tripathi, and S. Panchanathan, “Automated

gesture segmentation from dance sequences,” in Sixth

IEEE International Conference on Automatic Face and

Gesture Recognition, 2004. Proceedings, 2004, pp. 883

– 888.

Kavan, L., S. Dobbyn, S. Collins, J. \vZára, and C.

O’Sullivan, 2008. “Polypostors: 2D polygonal

impostors for 3D crowds,” in Proceedings of the 2008

symposium on Interactive 3D graphics and games, New

York, NY, USA, pp. 149–155.

Kyriakaki, G., Doulamis, A., Doulamis, N., Ioannides, M.,

Makantasis, K., Protopapadakis, E., Hadjiprocopis, A.,

Wenzel, K., Fritsch, D., Klein, M., Weinlinger, G.,

2014. "4D Reconstruction of Tangible Cultural

Heritage Objects from web-retrieved Images,"

International Journal of Heritage in Digital Era, vol. 3,

no. 2, pp. 431-452.

Laumond, and D. Thalmann, 2006. “Real-time navigating

crowds: scalable simulation and rendering,” Computer

Animation and Virtual Worlds, vol. 17, no. 3–4, pp.

445–455, 2006.

Li, R., T. Luo, and H. Zha, 2010. “3D Digitization and Its

Applications in Cultural Heritage,” in Digital Heritage,

M. Ioannides, D. Fellner, A. Georgopoulos, and D. G.

Hadjimitsis, Eds. Springer Berlin Heidelberg, pp. 381–

388.

Li, R. and S. Sclaroff, 2008. “Multi-scale 3D scene flow

from binocular stereo sequences,” Computer Vision

and Image Understanding, vol. 110, no. 1, pp. 75–90,

2008.

Linaza, M, Kieran Moran, Noel E. O’Connor, 2013.

"Traditional Sports and Games: A New Opportunity for

Personalized Access to Cultural Heritage", 6th

International Workshop on Personalized Access to

Cultural Heritage (PATCH 2013), Rome, Italy.

Matsuyama, T., X. Wu, T. Takai, and S. Nobuhara, 2004.

“Real-time 3D shape reconstruction, dynamic 3D mesh

deformation, and high fidelity visualization for 3D

video,” Computer Vision and Image Understanding,

vol. 96, no. 3, pp. 393–434.

Menier, C., E. Boyer, and B. Raffin, 2006. “3D Skeleton-

Based Body Pose Recovery,” in Proceedings of the

Third International Symposium on 3D Data Processing,

Visualization, and Transmission (3DPVT’06),

Washington, DC, USA, 2006, pp. 389–396.

Mitra, N.J, A. Nguyen, and L. Guibas, 2014. “Estimating

surface normal in noisy point cloud data,” International

Journal of Computational Geometry & Applications,

vol. 14, no. 04n05, pp. 261–276.

Moon, K. and V. Pavlovic, “3D Human Motion Tracking

Using Dynamic Probabilistic Latent Semantic

Analysis,” in Canadian Conference on Computer and

Robot Vision, 2008. CRV ’08, 2008, pp. 155 –162.

Pettré, P. de H. Ciechomski, J. Maïm, B. Yersin, J.-P.

Laumond, and D. Thalmann, 2006. “Real-time

navigating crowds: scalable simulation and rendering,”

Computer Animation and Virtual Worlds, vol. 17, no.

3–4, pp. 445–455.

Pforsich, J., 1977. “Handbook for Laban Movement

Analysis", New York: Janis Pforsich, 1977.

Pons, J.-P., R. Keriven, and O. Faugeras, 2006. “Multi-

View Stereo Reconstruction and Scene Flow

Estimation with a Global Image-Based Matching

Score,” International Journal of Computer Vision, vol.

72, no. 2, pp. 179–193, Jul. 2006.

Ruhnke, M., R. Kummerle, G. Grisetti, and W. Burgard,

2012. “Highly accurate 3D surface models by sparse

surface adjustment,” in 2012 IEEE International

Conference on Robotics and Automation (ICRA), 2012,

pp. 751 –757.

Rusu, R. B., N. Blodow, and M. Beetz, 2009. “Fast Point

Feature Histograms (FPFH) for 3D registration,” in

IEEE International Conference on Robotics and

Automation, 2009. ICRA ’09, 2009, pp. 3212 –3217.

Santos L. and J. Dias, “Motion Patterns: Signal

Interpretation towards the Laban Movement Analysis

Semantics,” in Technological Innovation for

Sustainability, vol. 349, L. M. Camarinha-Matos, Ed.

Berlin, Heidelberg: Springer Berlin Heidelberg, 2011,

pp. 333–340.

Sakashita, K., Y. Yagi, R. Sagawa, R. Furukawa, and H.

Kawasaki, 2011. “A System for Capturing Textured 3D

Shapes Based on One-Shot Grid Pattern with Multi-

band Camera and Infrared Projector,” in 2011

International Conference on 3D Imaging, Modeling,

Processing, Visualization and Transmission

(3DIMPVT), 2011, pp. 49 –56.

Tecchia, F., C. Loscos, and Y. Chrysanthou, “Image-based

crowd rendering,” IEEE Computer Graphics and

Applications, vol. 22, no. 2, pp. 36 –43, Apr. 2002.

Woo, W., J. Park, and Y. Iwadate, “Emotion analysis from

dance performance using time delay neural networks,”

in proc. of the JCIS-CVPRIP, 2000, pp. 374–377.

Yamasaki, T. and K. Aizawa, 2007. “Motion segmentation

and retrieval for 3D video based on modified shape

distribution,” EURASIP J. Appl. Signal Process., vol.

2007, no. 1, pp. 211–211, 2007.

Yang, Y., H. Leung, L. Yue, and L. Deng, 2010.

“Automatically Constructing a Compact Concept Map

of Dance Motion with Motion Captured Data,” in

Advances in Web-Based Learning – ICWL 2010, vol.

6483.