Embed Size (px)
Citation preview
Acquisition, Representation, Query and Analysis ofSpatial Data: A Demonstration 3D Digital Library
Jeremy RowePartnership for Research in StereoModeling/Information Technology
Arizona State UniversityTempe, Arizona 85287-0101
Anshuman RazdanPartnership for Research in Stereo
ModelingArizona State University
Tempe, Arizona 85287-2703480-965-5368
Arleyn SimonAssociate Professor ResearchDepartment of Anthropology
Arizona State UniversityTempe, Arizona 85287-2402
ABSTRACTThe increasing power of techniques to model complexgeometry and extract meaning from 3D informationcreate complex data that must be described, stored, anddisplayed to be useful to researchers. Responding to thel imitat ions of two-dimensional (2D) datarepresentations perceived by discipline scientists, thePartnership for Research in Spatial Modeling (PRISM)project at Arizona State University (ASU) developedmodeling and analytic tools that raise the level ofabstraction and add semantic value to 3D data. The goalsare to improve scientific communication, and to assistin generating new knowledge, particularly for naturalobjects whose asymmetry limit study using 2Drepresentations. The tools simplify analysis of surfaceand volume using curvature and topology to helpresearchers understand and interact with 3D data. Thetools produced automatically extract information aboutfeatures and regions of interest to researchers, calculatequantifiable, replicable metric data, and generatemetadata about the object being studied. To helpresearchers interact with the information, the projectdeveloped prototype interactive, sketch-based interfacesthat permit researchers to remotely search, identify andinteract with the detailed, highly accurate 3D models ofthe objects. The results support comparative analysis ofcontextual and spatial information, and extend researchabout asymmetric man-made and natural objects.
Categories and Subject DescriptorsH.3.7 Digital Libraries – Collections, Standards,Systems Issues
General TermsMeasurement, Documentation, Standardization.
KeywordsDigital Library, Geometric Modeling, ImageDatabases, Information Visualization, Physically
Based Modeling, Scientific Visualization, ShapeRecognition, WWW Applications.
1. INTRODUCTIONThe increasing power of computing, techniques tomodel complex geometry and compare to identifysimilarities has created powerful new capabilities toanalyze and interact with data representing three-dimensional (3D) objects. The techniques to model andextract meaning from 3D information create complexdata that must be described, stored, and displayed to beuseful to researchers. Responding to the limitations oftwo-dimensional (2D) data representations perceived byaffiliated discipline scientists, the Partnership forResearch in Spatial Modeling (PRISM) project atArizona State University (ASU) developed modeling andanalytic tools that raise the level of abstraction and addsemantic value to 3D data. The goals have been toimprove scientific communication, and to assist ingenerating new knowledge, particularly for naturalobjects whose asymmetry limit study using 2Drepresentations. The tools simplify analysis of surfaceand volume using curvature and topology to helpresearchers understand and interact with 3D data. Thetools produced automatically extract information aboutfeatures and regions of interest to researchers, calculatequantifiable, replicable metric data, and generatemetadata about the object being studied. To make thisinformation useful to researchers, the project developedprototype interactive, sketch-based interfaces thatpermit researchers to remotely search, identify andinteract with the detailed, highly accurate 3D models ofthe objects. The results support comparative analysis ofcontextual and spatial information, and extend researchabout asymmetric man-made and natural objects.
2. Background and PurposeDigital libraries are in the midst of a rapid andsignificant evolution. In less than two decades, thescope and complexity of digital collections haveblossomed from the CR-ROM and videodiscs of the firstAmerican Memory project to the myriad offeringsavailable via the Internet.
Similarly, scientific tools have evolved dramaticallyfrom observation, two-dimensional measurements, andstatistical computation to include complex three-dimensional models and visualizations even in
Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copiesare not made or distributed for profit or commercial advantage and thatcopies bear this notice and the full citation on the first page. To copyotherwise, or republish, to post on servers or to redistribute to lists,requires prior specific permission and/or a fee.JCDL Conference ’03, May 27 – 31, 2003, Houston , Texas.Copyright 2003 ACM 1-58113-000-0/00/0000…$5.00.
traditionally low technology disciplines such asanthropology.
Figure 1. 3D model of Hohokam ceramic vessel.
Digital libraries today offer text, graphics, images andincreasingly video, sound, animation, and sophisticatedvisual displays. Some now display three-dimensionalobjects, and permit the user to rotate and view an imageof the original object in their browser window usingQuickTime, plug-ins, or custom applications. A fewexamples include:
• -Forma Urbis Romae project at Stanford (PHP,MySQL and QSplat 3D viewer)
• -University of Texas Digimorph - CAT scan derivedsurface models of biological specimens displayedusing QuickTime
This paper presents an overview of the development andapplication of powerful tools for geometricvisualization and analysis used to create a 3D Digitallibrary to capture, analyze, query, and display three-dimensional data by the Partnership for Research inStereo Modeling (PRISM), an interdisciplinary researchteam at Arizona State University. Components of theprocesses include:
1. Metadata Schema and Organizational Structure
2. 3D data Acquisition - Scanning
3. Feature Extraction
4. Region Identification
5. Query Interface
6. Evaluation
As objects become more complex in terms of variety ofshape and changes in curvature, it becomes moredifficult to quantify and analyze. By developingmathematical techniques to represent the shape andcurvature, accurate models of the surface of 3D objectssuch as ceramic vessels, bones, or lithics can be created.These surface models and sophisticated mathematicaltools developed present the ability to analyze, identify,and compare the objects that they represent. The
accuracy of the measurements derived from the 3Dmodels created equal or exceed those possible usingtraditional 2D tools such as calipers and rulers. Inaddition, measurements such as height, width, maximumheight or width, surface area, or volume can be easily,consistently and accurately calculated, even forasymmetric natural objects.
Use of 3D data also makes possible new measures basedon topology and global or local changes in curvaturethat define the shape of the original object. The projectbuilt an interdisciplinary team of discipline andcomputer scientists, and technologists to guide aninteractive development processes. The disciplinescientists initially posed research questions, and thenthe computer scientists developed tools and spatialmodeling techniques to address them. Usingmathematical models and surface and volumeinformation, many new and powerful analytic toolsbecome available to spatially analyze objects. Forexample, boundaries between surfaces can beobjectively identified, small local areas of changes incurvature identified and compared, and accurate,replicable measurements calculated automatically.
Once meaning has been linked to the changes intopology, shape, or curvature by the domain scientists a“feature” is defined. The modeling process provides anobjective method to calculate physical measurements,and to identify boundaries and local areas of interest toresearchers by the changes that are associated with thefeature.
Figure 2. Components of interest for Ceramic Vessel
Once identified, each feature can be described by its size,position, shape or curvature. Examples of features thatcan be extracted from the model data include themaximum diameter or height of a ceramic vessel.
Features can also be components of interest to theresearcher that are mathematically abstract, such as thebase or neck of a vessel, keel of a ship, boundaries of thejoint surfaces on a bone or spindles that form in thenucleus of a cell during meiosis. Often the toolsdeveloped to identify features and regions also provideadditional capabilities that raise new research questionswithin the disciplines. These tools needed become newdesign challenges for the computer scientists, fosteringa new cycle of tool development. For example, ceramicanalysts have found tools that identify mathematicallydefined features found on the vertical profile curve of avessel such as end points, points of vertical tangency,
inflection points and corner points as features extremelyhelpful in analyzing vessel shape and style.
Figure 3. Laser scanning surface of ceramic vessel
In addition to the tangible research benefit the tools andtechniques provide, a significant result of this processhas been “cross-pollination” that has occurred asgraduate students and faculty from different disciplinesgravitate to a given project and explore application oftools and techniques to other discipline research.
A summary of data acquisition and analysis processesbegins with initial laser scanning to acquire the 3D datathat represents the object. Mathematical modeling i sthen applied to identify features and regions of interestto the domain scientists. Software tools developed bythe project team generate analytic data about theoriginal object, automatically assign metadata aboutspatial characteristics, and populate the database.
Figure 4. Point cloud of ceramic vesselcombined from multiple scans.
A visual query process was developed to permitresearchers to interact with the data using bothcontextual (text and numeric descriptive data) andspatial (shape and topological attribute) data. A sketch-based interface was developed that permits users toinput both context and sketches to visually describe theobject to initiate the search. Several text and spatialmatching algorithms are used to identify and rank orderobjects within the database that math the search criteria.
Initial development of the digital collections focusedon Classic Period (A. D. 1250 – 1450) prehistoricHohokam ceramic vessels from central Arizona housed
at the Archeological Research Institute at ASU.Additional development has involved bone shape andsurface, lithic tools, brain structures and DNA structuresin fertilized mouse egg cells. Research has extended toother disciplines with interest in spatial analysisincluding cloud formation, wind erosion, and facialrecognition.
3. Methods
3.1Metadata Schema and OrganizationalStructureOne of the greatest challenges in an interdisciplinaryresearch effort is coordinating expectations among teammembers, and developing communication processes thatbridge conceptual, strategic, and linguistic differencesacross the disciplines.
An iterative process was developed to share researchquestions, tools and intellectual approaches acrossdisciplines at project meetings. The results were agradual bonding of researchers, development of a sharedvocabulary, and substantial interaction about potentialresearch issues and approaches. These efforts provided afoundation for the initial modeling and analysis, and fordeveloping the metadata structure needed to organizedata for storage, analysis, and query.
A schema is an information class hierarchy that defines ashared vocabulary and a structure for documentsdescribed by that vocabulary. Use of a common schemafor all project data regardless of location offers thepotential to link and search across all of the databasesthat share the common schema. Dublin Core and Councilfor Preservation of Archeological Records (COPAR)metadata structures were used as foundations for schemadevelopment for this project. Extensible MarkupLanguage (XML) tags were used to describe thecontextual and spatial data elements, and for query anddisplay of data.
A conceptual goal of the metadata component of theproject was to develop an extensible schema structurethat could accommodate adding new types of objects asthe project continued to evolve. An object class wasdefined as the master class document type definition(DTD) for each item in the digital library database. Forthe 3DK digital library project, all of the additionaldescriptive data about each object was defined andorganized as contextual or spatial classes.
Contextual types define text and metric informationabout the object. This context class includes subclassesfor metadata associated with objects as they areacquired, processed, and archived such as type, itemname, catalog number, collection, provenance, etc. Atthis phase of the project these fields were primarilydetermined by existing descriptive data elements,though efforts were made to design a schema structurethat would accommodate adding new object types asnecessary. To date, several iterations to refine theschema model to function effectively across objecttypes have been completed.
Spatial data types define the 3D attributes of the object,including raw data, thumbnails, models, and calculated
or derived data about the topology, shape, andcomposition of the object. Use of common descriptivecomponents and geometric elements as new object typesare added will permit shared use of the modeling andanalysis tools across classes of objects. The project goalis to develop standards for description and organizationthat permit automated cataloging and population of dataas objects are scanned and processed for entry into thedatabase.
Due to familiarity and availability of resources, an SQLdatabase was used to store the contextual and spatialdata. Fields were assigned to each data element and largespatial data files were stored as hyperlinks. Generallyaccepted data formats such as binary, PLY, HTML, andXML have been used to make data accessible andsimplify migration and access to the data over time.
Figure 5. Point Cloud of scanned bone surfaceby Matt Tocheri.
3.2 Scanning and 3D Data AcquisitionThree-dimensional data about objects can be obtainedfrom many sources. Laser scanners can capture surfacedata for 3D objects ranging in size from a fewmillimeters to large buildings and bridges. Scanninglaser microscopes create three-dimensional data bystacking or “sandwiching” thin two-dimensional slicesof data. CAT scanners and MRI also capture 2D imagesor "slices" of a 3D object. Each of these devices has thecapacity to capture internal data about the objects inaddition to the surface outline in each slice. The 2Dimage slices can be stacked to assemble a 3D volumetricmodel of the object. These techniques are used to createthree-dimensional models of objects such as humanbodies, artifacts, and manufactured objects.Improvements in the resolution, portability, and cost aremaking three-dimensional data capture devices and thedata that they produce obtainable by many researchers.
The PRISM Digital Library project uses two Cyberwarescanners, the M15 and 3030 to scan ceramic vessels,bones, and other objects up to roughly a 30” maximumdimension. The object is scanned by a laser, whichcaptures spatial data (x, y, z) values for each point. Theobject is then rotated, and scanned again to captureadditional data. This process is repeated until sufficient
scans are obtained to combine to create a point cloudmodel to document the surface.
The Model 15 laser digitizer captures surface data pointsless than 300 microns (0.3mm) apart, producing high-density triangular meshes with an average resolution ofover 1000 points per cm2. The digitized data generatedby the scanner is composed of thousands of (x, y, z)coordinates that describe a point cloud that representsthe surface of the object scanned. Further analysisrequires generating a surface model from the pointcloud.
The simplest method to generate a surface thatapproximates the original smooth continuous objectfrom the thousands of points collected during scanningis to join adjacent data points to form a triangle mesh.The triangle mesh models the object surface, describingthe 3D object with both geometry and topology(Hamann et al., 97; Amreshet al, 2002). The geometrydescribes how the various points are distributed inspace while the topology describes the relationshipsbetween the points as they are connected in space toform the surface. IT is possible to take several snapshots around the object and “stitch” them together tocreate a displayable object using techniques such asQuickTime (QTVR). However, the QTVR representationis only a collection of 2D images, which can bepresented to emulate a 3D view, and permit usercontrolled zoom or rotation of the display. Though theimage appears to be 3D, there are significant differencesbetween the QTVR display and an actual 3Drepresentation of the object. The geometry and topologyconveyed in a true 3D representation offer the capacityto derive measurements such as volume, surface area,diameters, height, or distances between points.
Figure 6. Region editor applied to trapezium data modelby Matthew Tocheri
Modeling techniques are needed to create an actualmeasurable surface that represents the original object. Inaddition to the triangle meshes, PRISM software canrepresent these surfaces as Non-Uniform Rational B-spline (NURB) or subdivision surfaces (Bernadini et al.,98; Razdan et al., 98; Farin, 01, Farin, 02). NURBrepresentation provides the capability to assesscurvature distribution in complex objects such as
identification of the joint surfaces from scanned data ofa bone.
The accurate model of the object that results from thisprocess provides the data and conceptual frameworkneeded for objective, replicable analysis of surface andvolume attributes of the objects under study.
3.3 Feature ExtractionOnce the geometric structure has been obtained, the nextstep is to identify features and regions of interest to thediscipline researchers. Ceramicists look for shape,symmetry, and curvature, cellular biologists look forstructure of bio-molecular machines inside a cell,forensic anthropologists look at shape, and surfacecomparisons. A number of 3D modeling and analyticalgorithms have been combined, and new techniquesdeveloped to segment the geometric structure intoregions, and to identify meaningful features.
PRISM researchers developed a watershed-based hybridfeature extraction and segmentation scheme to workwith the triangular meshes (Mangan and Whitaker,1999; Pulla et al., 2002; Razdan and Bae, 2002). Thealgorithm automatically segments the surface intoregions of similarity based on curvature. The areasidentified can then be merged with adjacent similarregions, or split into smaller meaningful segmentsbased on threshold values that have been defined byresearchers. The nontrivial challenge has been totranslate these aspects important to the disciplinescientists into mathematically definable terms. For
example, the transition between a vessel neck and bodycan be described mathematically as an inflection point,and the maximum width of a vessel by diameter.
Interaction within the project team has resulted incrosswalks of definitions that help translate terms andpermit mapping mathematical concepts onto featuresmeaningful to the discipline scientists. The use of 3Ddata permits accurate identification of maximum andminimum measurements and calculation of complexmetric and descriptive data that are extremely difficultto obtain using 2D representations, linearmeasurements, and traditional measuring tools,particularly for naturally asymmetric or man madeobjects such as ceramics.
3.4 Region IdentificationThe second program developed is Region Editor thatcalculates more complex information about the objectand its component features such as total object volume,absolute object symmetry, the area of surfacesidentified, and the average angle at which surfacesintersect. Several of these measures are extremelydifficult to determine accurately using traditionaltechniques, particularly for asymmetrical objects. TheRegion Editor also permits researchers to add contextualinformation such as technical data about the scan, imageprocessing that has been used, provenance, or collectionto the 3D data. The final action of the Region Editor is tocreate the metadata or XML file associated with the 3Ddata for archiving.
Figure 7. Diagram of Visual Query Interface and Search Process
3.5 InterfaceFrom the perspective of interface, a primary designproblem was how to accept input for both contextualand spatial searches. An interdisciplinary “visual queryinterface” team guided research into interface design,identification of desired capabilities, development ofthe interface, and ongoing revision based on evaluationdata.
The PRISM team chose to design separate contextualand spatial input areas in the interface screen. Textualdata was input or selected from pull down menus toquery existing descriptive catalogs or databases. Searchcriteria include metadata such as name, type or numberof the item, collection, or other catalog informationabout the object. This input area also permits the user tolimit search by provenance by limiting the search to aspecific collection, or by measurements such as height,width, or maximum or minimum diameter.
Figure 8. Prototype profile-based visual query interfacefor searching ceramic vessels
The most interesting interface design challenge wasaccommodating the input to query spatial data and toidentify matching 3D shapes (Sakurai and Gossard, 88;Osada et al 2001, 2002; Razdan et al. 2001). To mirrorthe 2D profiles of the ceramic vessels familiar toanthropologists, the initial interface model used aninteractive vessel profile to define the spatial searchcomponent. A gridded area presents a sample of a profilecurve selected from the menu, or permits the researcherto draw a profile to be searched. Using the mouse andtool palette, the user can interactively create ormanipulate the shape until it represents the desiredvessel. Initially developed as a Netscape plug-in, thesketch interface has been converted into a Java applet tosupport multiple browsers and platforms.
After descriptive information about context and shapehas been entered, the query is submitted. The descriptiveand spatial information are separated and the multipledatabase queries are coordinated by project software.The contextual component of the query is handled as aconventional text and numeric database search. Thespatial search uses a variety of size, shape, and curvematching algorithms developed by the project team toidentify and locate similarities within the databases.
During search and analysis of potential matches,
intelligent filtering techniques are used limit the searchpool. Initially simple text, metric, or gross spatialclassification criteria are used to identify possiblematches from the database and reduce the search domain.As the search progresses, increasingly more complexalgorithms are applied to the shrinking pool ofpotential matches. The goal is to minimizecomputational load and search time while accuratelyidentifying all objects that match the search criteria.
Figure 9. Interface screen with ranked search results.
Another algorithm ranks the query results bydescriptive and spatial similarity to the query image.Query response information is presented sequentiallyover several screens, each providing an additional levelof information about the selected objects. The firstscreen displays thumbnail images and brief descriptionsof the top search results. Also presented is a large 3Ddisplay of the top search result, along with moredetailed descriptive and calculated information. The 3Dmodel can be displayed as a point cloud, wire frame, orfull shaded surface representation at the discretion ofthe researcher. Using the mouse, the model can berotated and viewed from any angle. Selecting athumbnail of another search result from the cue of searchresults will replace its model in the 3D display window.
If more detailed descriptive information is desired, athird window that displays the 3D model, and twoadditional analytic tools - a profile curve and curvatureplot, and additional descriptive data about the object. Afourth window can be selected to provide access to thecomplete descriptive and calculated data available.
Significant effort has been given to adapting theinterface design to accommodate the differences incontextual data and analytic tools between differentclasses of objects. The object type metadata can be usedto select the customized search template with fields forthe contextual and spatial data appropriate for theobject.
The visual query interface team developed trainingmaterials to guide new users and evaluation instrumentsto obtain formative guidance from users.
3.6 EvaluationSeveral techniques were used to evaluate and guide thedevelopment of the project. In addition to generalmeetings and team building activities, process mapping
and interviews of project team members providedqualitative and quantitative input to help buildcommunication among researchers in the team. Thisiterative process has extended throughout the project.
Initial evaluation input regarding interface componentsand design were obtained from the roughly 25 projectteam members. The current version of the interface wasused and assessed by the entire group at general andvisual query interface team meetings throughout itsdevelopment. The designs were critiqued, limitationsidentified, additional capabilities desired described,development challenges identified, and componentwork delegated to project teams.
Several evaluation sessions were held to obtain inputfrom faculty and student researchers outside of the team.After initial orientation, research problems were posedto the evaluation groups, and users used the interface tolocate individual target objects by context, shape orsize. Users were encouraged to explore the 150 ceramicvessels in the test database and comment on the clarity,scope, and ease of use of the interface. A revision cyclefollowed each evaluation.
4. FindingsThe spatial and volume modeling and analytic toolsdeveloped by the project team permit disciplineresearchers to quantify and accurately replicatemeasurements of complex 3D objects. The feature andregion recognition capabilities assist in visualizingcomplex, abstract concepts of interests to disciplineresearchers.
The iterative design process and team interactionevolved and worked well, particularly as the projectscope and development focused in the second year ofthe project. The regular interdisciplinary interactionamong the faculty and students was essential indeveloping comprehensive metadata schema, andprovided positive, constant pressure to extend theproject design requirements.
The challenges involved in developing the conceptualmodels to extend textual and metric contextual data anddevelop metadata for surface modeling were significant,but were exceeded by those that arose as volume datawas addressed. As the tools and techniques developedfor volume data became available, several newcapabilities became available to extend analysis ofsurface models. The growth of capabilities that resultedfrom this iterative process would have been virtuallyimpossible in traditional research that focused on asingle discipline.
The conceptual model developed to describe data usingobject class with subordinate context and spatialcharacteristics worked well to guide development ofboth analytic tools, and the query interface. The abilityfor users to simultaneously query by context and shapewas essential, and provided significant challenges forboth computer scientists and interface designers. Initialdevelopment of the sketch-based 2D profile model forthe spatial query laid the foundation for the even morecomplex development of the full 3D input modeling forquery input that is currently underway.
The capabilities and standards of virtually everytechnological component of the project were in fluxduring the course of the project. A few examplesinclude:
• -Portable scanners evolved considerably in terms ofaccuracy, rivaling the larger fixed scanners by themidpoint of the project.
• -Metadata and XML capabilities and standards wereevolving dramatically. Schema development toolsand strategies became more powerful.
• -Initial SQL search capabilities began to beaugmented by SOAP and XML search capabilities.
• -The reliability of distributed servers, databases,and networked access tools has evolvedsignificantly since the start of the project.
• -Java development permitted replacement ofNetscape plug-ins to simplify configuration andease cross platform access, and evolution of Java3D has provided significant new opportunities fordevelopment of a full 3D query interface.
Figure 10. Prototype bone joint surface tool interfaceby Matt Tocheri
5. DiscussionOne of the pleasant surprises during this project hasbeen the ease of extending the modeling and analytictools developed for one specific discipline to otherresearch domains and the interactive growth of the toolsfor surface and volume modeling and analysis. Theimprovements that have resulted from the iterativeprocess of identifying a domain research question,developing an application tool, deployment, analysis ofpotential applications across other research domains,and identification of new research questions hasgenerated significant process in developing modelingand analytic tools applicable to 3D data.
As 3D data acquisition tools become more affordableand readily available, the amount of 3D data that mustbe described, stored and displayed will growdramatically. Accommodating this huge datamanagement challenge will require development ofstandards and tools to begin to analyze and addmeaning to the data.
Several efforts are underway by the PRISM team, or are
planned to further extend the capabilities of the toolsdeveloped, and their application to domain research. Interms of infrastructure, the move from custom plug-insto Java will simplify deployment.
We are exploring alternatives to the SQL databasecurrently used, such as object-oriented databases.Another effort to improve searching is a pilot XMLsearch protocol developed by the National ScienceFoundation Biological Databases and Informaticsproject at Arizona State University (BDI) researchproject in conjunction with the ASU Long TermEcological Research (LTER) Metadata Committee andthe Knowledge Network for Biocomplexity (KNB)Project at the National Center for EnvironmentalAnalysis and Synthesis (NCEAS). The “Xanthoria”metadata query system Developed by this project teamuses SOAP (Simple Object Access Protocol) to sendXML query requests and responses, and supportssimultaneous web-based querying of distributed,structurally different metadata repositories.
The spatial analytic tools continue to develop asimprovements are made in the feature extraction andregion editing applications and more powerfultechniques are developed compare curvature, identifymatches and rank search results. Key to these efforts i sthe expanding partnerships with other research areaswith their own unique modeling and visualizationneeds. Included to date are more complex anatomicaldata from CAT scanners and MRI, cloud formationpattern recognition, geological erosion, andidentification of targets within complex, noisyenvironmental data.
Interface design continues to evolve. The project i sevaluating models developed for 3D query and displayby other projects including:
• -Princeton 3D Models Search Engine using TakeoIgarashi’s Teddy 3D sketch interface
• -National Center for Biotechnology Information(NCBI) Cn3D Genetic viewer
The development of a realistic 3D interface models thatpermit the researcher to sculpt the query in 3D space i sprogressing, as are additional analytic tools such asplanar overlays to visualize and objectively comparejoint surfaces of bones. Techniques to bookmarksearches to permit replication and simplify comparisonof objects within the databases are also being explored.A complex variation of bookmarks involves providing areplicable trail for researchers using the region editorand additional analytic tools such as the planar overlayto interact with the data and create their own interpretivemodels. Creating storage techniques for these derived,researcher defined or modeled data, and managing“version control” to permit replication anddeconstruction of the analysis is another challenge.
User evaluation of the current interface layout, colorpalette and design continues using both surface andvolume model data. In addition to initially developingspecific bone or ceramic vessel interfaces for thedifferent research domains, the project is working toidentify commonalities and conventions to develop aunified interface model. This common design appears tobe possible in initial query interface screens, with
differentiation of interface display occurring as objectsare identified, search results are returned, and researchersdrill down into object data that may vary acrossdisciplines.
6. ConclusionsDevelopment of the current model 3D digital library hasbeen an interesting exercise in interdisciplinary projectdevelopment. Translation between disciplines has takentime and effort. Even when common vocabulary is used,the discipline specific definitions and nuances can varysignificantly.
The spatial modeling tools developed to identifyfeatures and extract regions of interest have provenvaluable additions to research in the partneringdisciplines. The initial challenges have focused on dataacquisition, and development and display of models.
Initial digital library efforts to display images of surfacemodels using QuickTime and plug-ins havesignificantly expanded research and science educationas complex natural objects become approachablethrough such visualization. Adding modeling andanalytic tools based on surface and volume that permitobjective quantification and analysis of 3D data has thepotential to further extend research in virtually everydiscipline studying 3D objects.
As 3D data and the tools for visualization and analysisbecome more available, there is an increasing need forintuitive interfaces to provide gateways to the data.Researchers bring different strategies and approaches,and learning styles differ widely across potential usersof 3D data. Visual literacy and the sophistication ofusers also vary dramatically among users.
Standards are needed for data description, storage,interchange, and searching. Understanding of thiscomplex multidimensional data will be essential asrecords managers begin to interact with collections of3D data, and as it begins to reach archives. Conventionsfor display and organizing research tools are essential toeffective preservation and access.
Evaluation and continued research into learning styles,communication preferences, and visual communicationand display are needed to guide interface design.Clearly, development of simple, elegant, easy to useinterfaces to accommodate the range of tools and userpreferences for spatial data and modeling will be asignificant challenge now and in the future.
7. AcknowledgementsThis work was supported in part by the National ScienceFoundation (grant IIS-9980166) and funding from theVice Provost for Research and Economic Development atArizona State University. The authors would like tothank all of the collaborators that make up thePartnership for Research in Spatial Modeling (PRISM)team, particularly Anshuman Razdan, Gerald Farin,Daniel Collins, Peter McCartney, Matthew Tocheri, MaryZhu, Mark Henderson, Arleyn Simon, Mary Marzke,Myung Soo Bae, Matt Tocheri, David Van Alfen andDavid Capco. For more information on the 3DKnowledge project visit http://3dk.asu.edu.
7.6 ReferencesA. Amresh, G. Farin A. Razdan, Adaptive SubdivisionSchemes for Triangular Meshes, will appear in"Hierarchical and Geometric Methods in ScientificVisualization", edited by G. Farin, H. Hagen, B. Hamann,Springer-Verlag, 2002 (in print).
F. Bernadini, Mittleman, J., Rushmeier, H. Silva, C.,Taubin, G. The Ball-Pivoting Algorithm for SurfaceReconstruction, IEEE Transactions on Visualization andComputer Graphics, Vol. 5, No. 4, October/December1999.
Council for Preservation of Archeological Records(COPAR) http://copar.asu.edu/
Digital Morphology project at the University of Texashttp://www.digimorph.org/
D u b l i n C o r e M e t a d a t a Initiativehttp://dublincore.org/documents/2000/07/11/dcmes- qualifiers/
G. Farin “Curves and Surfaces for CAGD”, 5th ed.,Morgan-Kaufmann, 2001
G. Farin: History of Curves and Surfaces in CAGD. In:Handbook of CAGD, G. Farin, M.S. Kim, J. Hoschek(eds), Elsevier, 2002.
Forma Urbis Romae project a t Stanfordhttp://formaurbis.stanford.edu/docs/FURproject.html
B. Hamann, B. Jean, A. Razdan. CAGD Techniques in theControl of Surface Grid Generation. In: Thompson, J.F.,Weatherill, N.P., Soni, B.K. (eds.), Handbook of GridGeneration, CRC Press, Inc., Boca Raton, Fla. pp 29.1-26,1997.
K n o w l e d g e N e t w o r k f o r Biocomplexityhttp://knb.ecoinformatics.org/
Long Term Ecological Research project at Arizona StateUniversity http://caplter.asu.edu/
A. Mangan and R. Whitaker. Partitioning 3D SurfaceMeshes Using Watershed Segmentation. IEEETransactions on Visualization and Computer Graphics.Vol.5, No. 4, Oct-Dec 1999.
National Center for Biotechnology Information (NCBI)C n 3 D G e n e t i c viewerhttp://www.ncbi.nlm.nih.gov/Structure/CN3D/cn3d.shtml
National Center for Environmental Analysis andSynthesishttp://cochise.asu.edu/bdi/Subjects/Xanthoria/index.htm
R. Osada, T. Funkhouser, B. Chazelle, and David Dobkin,Shape Distributions , to appear in ACM Transactions onGraphics, 2001.
R. Osada, T. Funkhouser, B. Chazelle, and D. Dobkin,Matching 3D Models with Shape Distributions , ShapeModeling International, Genova, Italy, May 2001.
Princeton 3D Models Search Engine using TakeoIgarashi’s Teddy 3D sketch interface –http://www.cs.princeton.edu/gfx/proj/shape
Partnership for Research in Spatial Modeling at ArizonaState University http://3DK.ASU.EDU
A. Razdan and Myung Soo Bae, A Hybrid Approach toFeature Segmentation, in preparation.
A. Razdan, B. Steinberg, G. Farin. From Digitized Data toNURB Surface Meshes Proceedings of the InternationalConference of Rapid Prototyping and Manufacturing, pp749-754, Beijing, China,1998.
A. Razdan, D. Liu, M. Bae, M. Zhu, G. Farin, A. Simon, M.Henderson. Using Geometric Modeling for Archivingand Searching 3D Archaeological Vessels. CISST 2001June 25- 28, 2001, Las Vegas.
J. Rowe, Developing a 3D Digital Library for SpatialData: Issues Identified and Description of Prototype,R L G D i g i N e w s O c t o b e r 2 0 0 2 ,http://www.rlg.org/preserv/diginews/diginews6-5.html#feature1
H. Sakurai and D. Gossard, Shape Feature Recognitionfrom 3D Solid Models. In ASME Computers inEngineering, San Francisco, 1988.
U. Schurmans, A. Razdan, A. Simon, P. McCartney, M.Marzke, D. Van Alfen, G. Jones, J. Rowe, G. Farin, D.Collins, M. Zhu, D. Liu, and M. Bae, “Advances inGeometric Modeling and Feature Extraction on Pots,Rocks and Bones for Representation and Query via theInternet,” proceedings Computer Applications inArchaeology (CAA), 2001.
A. Simon, D. Van Alfen, A. Razdan, G. Farin, M. Bae, andJ. Rowe, “3D Modeling for Analysis and Archiving ofCeramic Vessel Morphology: A Case Study from theAmerican Southwest,” Proceedings of the 33rdInternat ional Symposium o n Archaeometry.Geoarchaeological and Bioarchaeological Studies,Vrije Universiteit, Amsterdam, 2002.
