GIRAPIM: A 3D INFORMATION SYSTEM FOR SURVEYING CULTURAL HERITAGE ENVIRONMENTS

Abstract. The information of public buildings, specially cultural heritage, is intended to be published and shared between citizens, rms and institutions. Usually, conventional GIS and CAD solutions do not include semantic information and common metadata, preventing application interoperability and the public access to information through Web Services. Moreover, rms demand support for entire building life-cycle requiring a high integration between applications and services. Following our proposal, the integration is decomposed into three large blocks for related three main tasks: documentation, information and management, sharing a common set of data and software tools for processing and analysis. Enabling advanced applications and including the interoperability between embedded applications and blocks demand a common vocabulary, managed by an ontology which is referenced to a common model framework, given by and adaptation of CityGML. In this paper we show our approach, including the design and implementation of GIRAPIM, a software tool enabling partial building modeling, semantic tag documentation and management of intervention processes in Cultural Heritage environments.

Introduction

Cultural Heritage environments, or more general built urban environments, must be considered as living systems where dierent kinds of agents interact between them by exchanging physical resources, functions and services to dierent scales. Their character as a living system involves to the whole life cycle relative to different stages and their corresponding tasks such as surveying (relative to land, urban, district and building objects), planning (relational database, updating), intervention (conservation or rehabilitation, usually), and maintenance (monitoring, tracking and validation). There are dierent kinds of agents which operate to dierent scales going from microscopical agents for environmental issues, human agents (understood as users or technicians with dierent kinds of interaction involving to accessibility or interventions, e.g.), physical domain (buildings, urban district and surrounding nature) or variable environmental conditions (atmospheric conditions, unexpected events). Following an obvious linguistic analogy with an increasing complexity, the corresponding space-temporal models can be described in syntactic (in terms of basic components), structural (elements, relations and basic laws for autonomous

2 subsystems) or functional framework (with tasks dened as paths, scalar or vector functions, dynamical systems or functionals to be optimized). Each one of them has a logical structure with the corresponding computer implementation in terms of classes, propositional, and descriptive logic. The most complex approach concerns to descriptive logic (used in ordinary language), not only because it extends the two precedents, but because it must incorporate patterns for interpreting the current information and generating new knowledge arising from the evaluation of functionals dened between components or subsystems. The unequal exchange between (physical or logical) agents holds in terms of unbalanced interrelations in live environments with a lot of problems involving urban ecosystems, and having strong and perhaps irreversible consequences on environmental conditions. The exchange between subsystems involves to the functional approach. Thus, a functional modeling of urban ecosystem is not only a theoretical problem, involving the coupling between competitive or cooperative subsystems: it involves tasks (to be described as functional solutions) associated to balanced exchanges between dierent daily aspects such as quality of life, sustainable use of natural resources and technologies. In particular, it is necessary to identify, measure, model, simulate and visualize interaction patterns in urban domains. The wide catalogue of modeling techniques, strategies, services and products includes: 1) multiagent systems (for modeling behaviors and interaction in a common environment), 2) eco-ecient strategies (economic and ecological for sustainability), 3) Services Oriented Architectures (SOA as computational framework for providing services) and 4) global monitoring (ubiquitous sensing and computing technologies as support for capturing, measuring and updating information) in increasingly networked scenarios. Unfortunately, our knowledge is still very fragmented. Currently, we have only models for describing mutual interaction, involving only to some partial aspects involving agents in very restricted physical domains. 1.1 Information Systems in AEC environments

Our multi-system approach includes three systems involving Documentation, Information and Management System which are superimposed to the initial dierent tasks (surveying, planning, intervention, and maintenance) like it is showed in gure 1. The above three Systems can be considered as very large layers which are superimposed as successive steps with an increasing complexity; in particular, Documentation System provides the support for Information System, and this one for Management System, with their corresponding geodatabase for storing, updating, processing and analysing information which must be translated in new knowledge about the environment. This approach aims to improve traditional Architecture, Engineering, Construction (AEC) solutions for Building Information modeling (BIM) (see [3]). Traditional three-dimensional Geographic Information Systems (3D GIS in the successive) include two kinds of data, where raster information is geometrically referred and managed in terms of vector data. The design and implementa-

GIRAPIM, a 3D Information System

tion of multiple articulations between the above three systems (Documentation, Information and Management) requires a semantic approach for the knowledge management. A semantic approach provides a support for interactions between dierent agents in terms of ordinary language with keywords which are structured according to acknowledged denitions (thesauri) and logical rules (taxonomies).

Fig. 1. Overview of systems and AEC task

CityGML [7, 9] provides the framework for developing applications focused towards planning and executing (conservation or rehabilitation) interventions in Cultural Heritage buildings or to provide services. CityGML is the new encoding standard, approved by the Open Geospatial Consortium in August, 2008 [4]. In particular, we aim to provide a support for solving physical and digital accessibility issues to dierent kinds of users, with a special regard to technicians (in charge of solutions) and disabled persons in very restricted environments relative to indoor (Museums or Cathedrals, e.g.) and outdoor urban scenarios (urban district of Segovia, Spain). 1.2 Information Systems for Cultural Heritage

In view of dierent requirements posed by dierent kinds of users (technicians, citizens, visitors with some dependence or disability, e.g.) with dierent requirements to perform planned tasks, we have adopted a multi-resolution approach based on a fusion of image- and rang-based 3D modeling. For instance, (a) high-level resolution is adapted to oce work requirements for tasks to be performed by technicians; (b) mid-level resolution is displayed on a mobile PC (including UMPC on wheelchairs for disabled persons, e.g.) for on-line updating information which are referred to high-resolution 3D models or for making easier customized queries (including access to the available database) or services (localization, additional information); (c) very low-level resolution is displayed

4 on tactile interfaces of small mobile devices (smart phones or mobile internet devices) with more constrained requirements which are linked to an interactive dialog with a monitored environment. Cultural Heritage environments where we have performed integrated solutions are given by two kinds of communication networks for indoor scenes (WiFi for the Maritime Museum in Barcelona) or outdoor scenes (Bluetooth or BT, for a historic urban district in Segovia). Each one of them has a localization system which is based in local integrated sensor networks for indoor scenes (radio-frequency identiers or RFID for Museums) or outdoor scenes (Global Positioning Systems or GPS for urban districts). The available communication network restricts the type of dialog, and the possibility of linked services; so, BT supports only a low-level interactivity which is based on deployment of multimedia contents, currently. The large diversity of hardware (proprietary solutions, electronic devices, etc), software (operating systems, visualization systems) and communication protocols, requires to solve interoperability issues according to standards. Solution adopted in PATRAC project follows a SOA architecture with services for monitoring, database management, information treatment (processing and analysis), and communication, where the only technical requirement for accessing is connected to a external local (Intranet) or global (Internet) network. Obviously, the mid-level CityGML functionalities can be used only at oce or on a laptop PC at workplace. However, we have developed some tools for a drastic reduction of range-based models which can be visualized on the display of small devices. The availability of localization systems in indoor or outdoor environments and the interactive visualization of 3D contents in Collada format [2] allows to make queries, update information and develop a dialog between technicians at workplace and oce. Moreover, we take advantage of support provided by Google Inc. web applications for providing additional contents which are visualized in the display of mobile device, and which will be activated following passive (automatic detection of presence by RFID tags) or active (tactile display to be navigated through user interaction). The rest of the paper is structured as follows. Section 2 describes how to extract geometrically accurate documentation for building model. In section 3 we explain the relevance of the Ontology that provides the Semantic Framework and extents CityGML features. In section 4 we summarize our low resolution hybrid modeling approach to get building models. Section 5 summarizes details of the CityGML extension designed to document accessibility problems in cultural heritage environments. The section 6 establishes bases of GIRAPIM software architecture and describe the eorts in order to achieve interoperability with current commercial applications. Finally section 7 shows the conclusions extracted from this work and future developments.

GIRAPIM, a 3D Information System

Reference Geometric Framework

Urban or architectural environment, spatial planning and interactions simulation in complex environments have requirements that aect to an increasingly complex knowledge representation with respect to a well-dened and robust model. The robustness is linked with geometric models: a well-dened geometry for an architectural context provides support for advanced visualization including features such as navigation, inspection and extraction of metric information and interactive simulation of possible interventions. Simultaneously, the information exchange between dierent experts and the implicit information reuse in 3D models requires solving interoperability issues between dierent sources. 2.1 3D Laser surveying

We use a hybrid approach to 3D architectural surveying arising from image- or range-based information, with georeferencing of capture devices in terms of UTM coordinates, including a topographic network, depending on customer requirements. In our case, 3D modeling of isolated buildings is centered at object; the right georeferencing allows to provide services in terms of the current localization of users. The range information of our 3D models for isolated buildings or small urban districts has been captured with dierent laser scan devices (see [10]); these devices produce dense clouds points which store geometric information (spatial coordinates for the position) and radiometric (one channel for intensity in grayscale or three channels for color information). Aligning dense point clouds requires the selection of homologous points in dierent scans, scanning and alignment of views (with special regard to problems of relative orientation and height), cleaning of redundant information (dierent criteria optimization) and simplifying information. Typically, laser scanning is used for surveying individual buildings, but has also been used for scanning small urban areas (squares, streets, districts) where private or public agents are planning to perform some kind of intervention; nominal reaching of laser scan device (Ilris 3D, Optech Inc.) is superior to one thousand meters. Some quarters of small villages have been scanned, by integrating cadastral information in the resulting 3D GIS, with some applications developed for local administrations. Scanning small urban areas generates clouds of tens or hundreds of millions of points that requires high computing power for advanced visualization with dierent levels of detail to include rendering, browsing, query, annotation and extraction of metric information. There are technical solutions via hardware (powerful machines) or software (virtual parallel machines) for intelligent compression or mesh simplication, but it requires the availability of solutions for desktop computers and, soon, for portable devices such as PocketPC, SmartPhones or UMPC that provide navigational tools, expert assessment or other services to citizens. A crucial fact is the availability of models at dierent resolution levels compatible with each other (common elements at each level) and tools for the progressive reduction of associated mesh to maintain the overall

6 topology of the object. Thus, system will be able to discriminate level to oer depending on the device prole that aims to access the application remotely. 2.2 Dominant planes detection

The specication of a semantic geometrically referenced framework facilitates the reuse of materials from a unique ontology with dierent knowledge elds involving the domain (physical domain, monitoring systems, equipments), users (technicians, citizens in general, dependent persons) and tasks to be performed (interventions, visits, interactive dialog with the environment). The geometrical support is given by multi-resolution models supporting dierent functionalities or tasks which are topologically dened as paths in the physical representation of urban space or as temporal agenda in the representation of management system. These models are geometrically dened by using geometric features common to dierent levels of detail, following CityGML methodology for urban environments or buildings. The triangular meshes are very accurate but they present additional memory requirements and a very high processing power. Otherwise, quadrangular meshes provide a support for rendering on low-cost portable devices and, therefore, monitoring of processes linked to interventions and the provision of services based on highly simplied models.

Fig. 2. Example of semi-automatic dominant planes detection

An extreme case of quadrangular mesh is linked to a 3D object corresponding to the collection of dominant planes (faades, roofs and oor) of a building which is represented as simple polyhedron (connected object without self-intersections).

GIRAPIM, a 3D Information System

A simple triangulation associated with the sample cloud points provides facets for small triangular faces. Each triangular facet is the support of a plane. A plane is determined by three non-aligned points that are ordered following an cyclically counterclockwise ordering, the vector product of vectors generated by two vectors that connect one point with the other two, provides the normal vector to the plane. There are dierent algorithms types (adjacency, triangle soup) that allow spread triangulation when the dierence of unit normal vectors is below a threshold (i.e. the dierence between axes in the unit normal sphere). The result of this process generates a collection of regions Ri. which is the maximal connected region contained in a plane with a selected normal unit vector; maximality is dened with respect to the aggregation of adjacent triangles with a similar unit normal vector. An automatic selection of modes (values with the highest frequency from statistical viewpoint) allow to identify the winner planes. The boundary of each planar maximal region is a non necessarily simple polygonal because it can contain holes (in other words, it is not necessarily simply connected); in the simplest case, this regions it is a rectangle providing support for matching views as maps of textures. The spatial intersection of adjacent dominant planes gives dominant edges, representing an extrusion of corners appearing usually in cartographic representations. The semi-automatic detection of dominant planes in GIRAPIM application requires to identify several parameters by the user such as thresholds for maximal dierence of length between axes of vectors or the maximal number of dominant planes for the scene. However, after xing meaningful parameters the detection and labeling of dominant planes is performed in an automatic way, by inserting typology (faade, roof or oor) in order to facilitate the automatic collections of structural elements generation connected by the condition of sharing a common edge. Finally, we obtain a reference model (see gure 2) for supporting vector information (see more in section 4).

Cultural Heritage Ontology

According to T. R. Gruber, an ontology is a specication of a conceptualization [5]. More explicitly, an ontology must provide a logical representation of relations between lexicon (vocabulary given by key words for the specic knowledge domain), thesauri (dictionaries) and taxonomies (systems of logical rules) with the corresponding owcharts. Entity-relationships provide a forma representation of such owcharts. Obviously, the articulation between components and the specied ontology are not unique. However, to avoid interpretation problems linked to interoperability between dierent Ontologies, a unique Ontology can be developed in the context of Cultural Interest Goods context (BIC in Spanish). This solution integrates aspects relative to domain (physical space), users (experts, disabled citizens and others) and tasks to be performed. Componentes are articulated between them in a dynamic way by using events that occur in each context, even though the meaning may be dierent. For instance, a ladder

8 (in domain component) can be linked to a structural analysis (related or not with an intervention) or a solution to an accessibility problem (for visitors with physical disabilities within the user component) in relation to tasks location or movement (included within the task component). Taxonomies are responsible for the logical rules management contained in relationships. There are dierent types of logical rules depending on the explicit expression of terms which are included in the vocabulary, the glossary or the relationship. They follow an increasing order of diculty in its formulation (logic of classes, logic of predicates and descriptive logic) and methodologies involving rst-order propositional logic, second order or fuzzy logic, e.g.. The machines usually operated with logic or propositional logic class, instead of humans in daily task, which operate as descriptive logic schemes in which the experience has a similar value or higher than the conventional logical schema. Therefore, to improve the knowledge management is necessary to design and implement the mechanisms to connect with descriptive formal logic. This goal is not merely academic, since it aects the improvement of communication between man and machines. Even fuzzy logic, initially applied in Robotics and developed for over twenty years in many dierent contexts, do not provide support for this kind of relations. For an eective management, it is necessary characterize and implement the Ontology architecture (lexicon, thesauri and taxonomies), populate and validate it. The rst stage requires the participation of experts from dierent areas (Architecture, Linguistics, Computer Science among others), while the last two, require the participation of rms, institutions related to BICs and end users. Therefore, we have developed a collaborative framework for the design and implementation of software architecture that manages assets and services related to use cases. The geometric support provides a common reference framework (see section 2) for information contained in relational databases. Our framework provides interoperability between additional tools for processing and analysis modules of the information contained in dierent repositories. Thus, metadata provides a lexicon support for accessibility and simplication of procedures for searching contents via web. However, they are not enough by themselves in order to allow interoperability: it is necessary to specify relationships aecting entities contained in dierent multimedia (text, image, volumes) resources in terms of keywords. Each set of keywords conforms a lexicon whose elements are represented as nodes of a graph; denitions of keywords stored in thesauri or glossaries, give rise to dierent relations which are represented as edges of a directed graph. Additional attributes (weights, paths, subgraphs) for supporting advanced knowledge management are being implemented in a complementary work. In this work, we aim to design and implement a software platform called GIRAPIM that provides a support for intervention processes to administrative entities, rms (SMEs in charge of services, usually), and/or cultural foundations. In this way, one has a common framework for dierent tasks involving 3D docu-

GIRAPIM, a 3D Information System

mentation (surveying, mainly) and linked Information and Management Systems for Cultural Heritage in urban environments. The resulting common platform can also be used for more specic tasks such as those related to accessibility problems for dependent persons with psychic or physical disability (motion, hearing or visual).

3D Information modeling

Information Processing and Analysis are the two rst stages of 3D building modeling for small urban zones. In our approach we have developed modeling techniques based on hybrid modeling (image- and range-based), CAD modeling (see gure 3) and rendering tools for visualization. Their hybrid character concerns not only to the dierent nature of inputs (rectied images and clouds of points), but also to the reconversion of discrete digital information to a continuous one: clouds of points are grouped in dominant planes which are grouped again in polytopes dened by connected polygonal of dominant planes with adjacency restrictions. This information is symbolically represented in dual terms for making easier their management from the computational viewpoint, as usual. In our case, the automatic detection of 3D primitive is performed from discrete information (dense points cloud). This detection follows a similar approach to the 2D case: ltering, local analysis and clustering, of PL-structures which are superimposed to the clouds of points. Clustering strategies have two critical parameters: linearity and proximity of unitary normal vectors. The resulting object is called an urban polytope corresponding typically to a block bounded by a collection of streets to the ground level, with true faades. Instead of extruding faades from aerial photography and cadastral information, and matching views, we perform a direct reprojection of views on the metric 3D model, which has been previously reoriented to minimize distortions . Next, we generate an urban model volumetric segmentation with the polytope, i.e. the decomposition of 3D scene in a nite union of isolated urban polytopes which are our intermediate cells. The subsequent urban model completion and customization is delegated to specic-purpose CAD modeling tools such as Google Inc. SketchUp or Blender by exporting mesh to DXF and Collada formats. The next step concerns to service provision related to Information Systems for AEC environments. Traditional powerful CAD modeling tools have not incorporated semantic contents allowed by CityGML, like Kolbe et Al. shown in [8]. Knowledge management and semantic interoperability are tasks delegated to ontology in an upper layer. In this way, each geometric object supports semantic information which can be independently collected for surveying or reporting specic issues regarding in our case to Accessibility (for citizens) or Rehabilitation and Conservation (for technicians) tasks in urban environments (see Fig. 4). The generation of complex and detailed building models is not an issue managed by GIRAPIM because it is a well resolved problem. However, GIRAPIM allows managing of dierent levels of complexity at dierent levels of detail (LOD) suitable to user application domain, based on CityGML idea. This multi

10

Fig. 3. CAD model for cultural heritage documentation

resolution approach allows to manage models at dierent resolutions with levels of complexity going from a simple cube model based on the automatic detection of dominant planes, going from simple building objects to more complex ones, such as a cathedral model given by a low-level triangular complex mesh. Hence, it provides a model to show urban scale contents and for annotation of detailed information, including point of interest or material information.

Heritage and urban documentation

GIRAPIM has been developed as collaborative environment for interactive 3D cultural heritage documentation. An important issue concerns to the application of the semantic support for 3D urban objects 3D Information System which is provided by CityGML. GIRAPIM develops this framework, by providing an extension of CityGML data model for cultural heritage environments based on thesaurus included in our ontology. Extending this framework is an aordable task thanks to CityGML Application Domain Extension mechanism. In our approach, we extend CityGML building module, adding three new modules according to the three functional components of Ontology (users, tasks and domain), creating the Cultural Heritage ADE (see gure 4 below). Each module can be described as follows: 1. Cultural Heritage. It is the main module that imports CityGML building module. For us, a building is a collection of spaces (not only a room, a broader concept). Also, each space can contain a set of structural elements (like walls, arches, doors, ...). These characteristics are also dened by CityGML as detailed below:

GIRAPIM, a 3D Information System

11

Fig. 4. CityGML ADE for documenting cultural heritage

Space = Room Constructive element = IntBuildingInstallation Monument (new entity) is a type of AbstractBuilding The Cultural Heritage ADE adds to CityGML some new entities (e.g. monument) and necessary attributes dened by the ontology and specied in thesaurus. 2. Pathology. This is a extension for the building module to identify accessibility problems and pathologies in buildings. An accessibility problem is dened by the space that contains it, while a pathology is associated with the construction element that suers it. 3. Intervention. This module includes new entities that represent the solution to solve pathologies and accessibility problems.

12 The Cultural Heritage ADE is a rst step in order to provides future services such as semantic support for monitoring, context-aware information, automatic reports generation or management agenda for intervention process (typically in 4D models).

GIRAPIM software architecture

Fig. 5. Overview of GIRAPIM software architecture

Technicians working in Cultural Heritage domains need to have an accurate model for information management. Furthermore, lower resolution models are desirable to provide an integrated system for documentation and information. For making easier communications between work at place and oce, it is necessary to share low-resolution models which can be updated by technicians and/or explored by citizens in dierent kind of devices like UMPC (Ultra Mobile PCs), Smart Phones or Mobile Internet Devices, in general. This goal motivates the design and implementation of our software application GIRAPIM as a client desktop tool. Figure 5 shows some connections between dierent components, with emphasis on the viewer, the repository and CityGML module. Next, we shall will briey describe each part of them: 1. Viewer. It is the component that enables visualization of semantic annotations and geometry 3D model. Based on OpenGL rendering technology (a Khronos Group standard), it shows the information overlaying a geometry

GIRAPIM, a 3D Information System

13

and semantic layer. It is also responsible for picking and selection of objects in the scene with the mouse. 2. Repository. It is designed to provide support to CityGML and Ontology thesauri classes. It manages the central repository of contents for scenarios where accessibility problems are detected. Information is managed and showed by GIRAPIM, but it is available for other external services and applications in the semantic database supported by PostgreSQL [11] database management system with PostGIS [12] extension enabled. 3. CityGML. This is the module responsible for Cultural Heritage ADE data exportation. The set of CityGML entities extended by Cultural Heritage ADE cover completely the ontology thesauri so database information contents can be exported and managed in a Web 3.0 way, allowing interoperability with other services (see gure 4). 6.1 Enabling interoperability

The main issue to solve in order to achieve semantic annotation simplication is the model interoperability. Unfortunately, we have not still an ontology allowing modeling representation with the advanced required functionalities linked with Management Systems for 3D modeling. Because of this, we have used the open standard Collada, which is compatible with many visualization and modeling available tools, such as Google Earth or Blender respectively. Collada (COLLAborative Design Activity) [1] is an open interchange le format for interactive 3D graphics, currently managed by the Khronos Group consortium. It denes a XML schema and vocabulary for exchanging digital graphics between applications. Originally created by Sony, today is shared between Sony and Khronos Group and it is supported by many game studios, graphic engines and modeling software (e.g. 3D Studio, Maya or Blender). GIRAPIM allows importing and visualization of Collada models (.dae) representing cultural heritage assets with up to high level of detail, especially at interior. GIRAPIM also support Google Earth (.kmz) le import feature. KMZ is an acronym for KML [13] compressed that can be easily linked with Collada graphic models. As an additional superimposed layer for these models, the users can include semantic tags displayable and navigable in 3D virtual globes and exportable to CityGML. For instance, data can be visualized in Google Earth, or in any web browser thanks to available plug-in. Thus, Collada simplies the access and visualization of the semantic information from future Web Services, as in [6]. Moreover, KMZ les can be easily exported from free version of Google Sketch Up allowing 3D surveying and reconstruction of buildings from satellitar images. Semantic annotation translates 3D geometric model data to 3D semantic model data in CityGML, which is required for agent reasoning capabilities. This approach takes advantage of CityGML geographic position of the semantic information to support advanced tasks for the development of Smart Cities such as spatial reasoning, problem detection or 3D context-based information provision. With the geometrically referenced information we can show context-aware

14 information in 2D maps for mobile devices, allowing intelligent remote content provision. The creation of Web Services to transfer 3D semantic information between many dierent proposal systems is supported in our framework. GIRAPIM provides data access and management to data for the reasoning and services tasks laying in an Ontology and CityGML common vocabulary. 6.2 Discussion

GIRAPIM allows semi-automatic low resolution building modeling; it is connected with specic software tools developed by our team involving the fusion of image- and range-based information, and linked Information Systems. Additionally, these tasks can be performed with professional tools designed for warranting interoperability and sharing methodologies with other agents. The lowest resolution services of GIRAPIM supports the interoperability with additional functionalities based on mobile services (including Google-based applications), provided an Internet access be available. From a more technological viewpoint, we are working in the development of an integrated complete city view, besides KML export feature, and a complete support and interoperability with BIM processes through IFC data. Following the same philosophy, we aim to include RDF publication of CityGML building data, in fully compatible way with Web 3.0, allowing an interactive dialogue between smart cities and users according to the character of cities as living systems.

Conclusions and future work

Urban GIS are emerging as a new way for planning, publishing and managing building information by institutions and citizens. The addition of semantic data to geometry and geography support enables application interoperability and creates a common framework to develop a network of Web Services that connect people, things and local administrations. The design and implementation of this kind of frameworks, open a common space for collaborative environments between citizens and the city as a whole. To achieve these goals it is necessary to go further in developing modules of integrated frameworks allowing to create, annotate and manage a georreferenced information. GIRAPIM is a prototype that support these three tasks under the common CityGML framework, and provides a support for low-resolution mobile services linked to Documentation, Information and Management Systems. Our solutions have been applied to indoor and outdoor scenarios in Spain including the Maritime Museum of Barcelona and historic urban city of Segovia, which are meaningful to provide services in 3D Georreferenced Systems. The linked mobile solutions are designed as web services, and because of this they are independent of proprietary solutions; they need only facilities for communications and network access (Intranet for the Museum, and Internet for the historic urban center of Segovia). The most meaningful case for urban environments concerns to a small zone of historic center of Segovia, where an integrated Services

GIRAPIM, a 3D Information System

15

Oriented Architecture has been developed to provide a SmartPhone-based support for persons with a motor disability. The developed SOA provides a support for dynamic interaction in urban complex environment including a customized provision of services (guide, visualization of multimedia contents, additional external services) from the module of localization. Customization depends on user preferences, and specications relative to his/her mobility capabilities, due to the accessibility problems in the historic center of the old city. Additionally, mobile services provide a support at workplace for making consults, navigating low-resolution models, visualize and update information in dierent multimedia formats, on smart mobile devices for technicians. We have already developed an Ontology for tasks to be developed by technicians, which is related to the application of non-destructive and semi-destructive techniques in conservation and restoration techniques. These tasks are included as modules in the Ontology to allow their management in a common framework. In the next future, they will be incorporated to our application to advance towards an eective integration of Documentation, Information and Management Systems.

References
1. R. Arnaud and M.C. Barnes, COLLADA: sailing the gulf of 3D digital content creation, AK Peters, Ltd., 2006. 2. M. Barnes, Collada, ACM SIGGRAPH Courses, ACM, 2006, p. 8. 3. D.A. Campbell, Building information modeling: the Web3D application for AEC, Proceedings of the twelfth international conference on 3D web technology, ACM, 2007, p. 176. 4. G. Groger, T.H. Kolbe, A. Czerwinski, and C. Nagel, OpenGIS City Geography Markup Language (CityGML) Encoding Standard, Open Geospatial Consortium (2008). 5. T.R. Gruber et al., A translation approach to portable ontology specications, Knowledge acquisition 5 (1993), 199199. 6. K. Honda, N.D. Hung, and H. Shimamura, Linking OGC web services to Google Earth, SICE-ICASE, 2006. International Joint Conference, 2006, pp. 48364839. 7. TH Kolbe and G. Groger, Towards unied 3D city models, Challenges in Geospatial Analysis, Integration and Visualization II. Proc. of Joint ISPRS Workshop, Stuttgart, 2003. 8. T.H. Kolbe, G. Groger, and L. Pliimer, CityGML-3D city models and their potential for emergency response, Geospatial Information Technology for Emergency Response (2008), 257. 9. T.H. Kolbe, G. Groger, and L. Plumer, CityGML: Interoperable access to 3D city models, Geo-information for disaster management (2005), 883899. 10. J.J.Fernandez L.M.Fuentes, J.Finat, Using Laser Scanning for 3D urban modeling, Urban and Regional Data Management UDMS 2006 Annual (2006). 11. B. Momjian, PostgreSQL: introduction and concepts, Addison-Wesley, 2001. 12. P. Ramsey, PostGIS manual, Refractions Research Inc (2005). 13. T. Wilson, OGC KML, version 2.2.0, Open Geospatial Consortium (2008).