Multimedia databases and infrastructures have gained tremendous attention in many application areas for the payoff in large interdisciplinary projects like e-commerce, digital libraries, medical information systems, or geographic information systems. Within the next years, a high industrial impact of multimedia databases is expected, and multimedia processing is expected to become an emerging topic in business processes that consider themselves knowledge enterprises. Traditional relational database systems store structured data like numbers and character fields adequately. Today’s information systems, however, need to handle complex multimedia data types in addition. The extension of traditional database systems toward multimedia database systems is a tedious task and we explore how commercial database vendors have succeeded in extending the functionality for multimedia data accessibility, speed of retrieval, and searching flexibility. Relational databases have served well and probably will continue to dominate the database market for quite a few years. But the focus is swinging toward object-oriented database technology as a new fundamental shift, last but not least caused by the emergence of the object-oriented Java programming language.

Today’s leading commercial database vendors offer substantial object-relational database extensions for processing multimedia data. In the first place, the two database systems IBM DB2 Database and Oracle 9i Database are to be mentioned which offer high-level multimedia support by dedicated object-relational functionality and multimedia packages. Furthermore, a variety of third-party and open source products exist which offer multimedia database extensions.

The question now is whether these multimedia database systems have reached a stage of maturation that qualifies them for operational use yet. We have tried to find some arguments for answering this question in the context of DaimlerChrysler’s specific needs. This document therefore summarizes test and evaluation results regarding multimedia capabilities of the following four commercial database systems: IBM DB2 (UDB), Oracle, Microsoft SQL Server, and Informix. These systems have been selected as they are most commonly used within the group. Various aspects such as functionality, performance, usability, as well as future development plans have been evaluated and to some extend been tested. We evaluated multimedia extensions mainly for image data (and partly for text data) as such data seem to provide the greatest potential for multimedia applications.

From our tests we expect costs for constructing and maintaining multimedia databases to be not significantly higher than for traditional databases. The benefit of using multimedia database we see in a greater flexibility for supported data types, query types and the basic architecture. The technical support offered for commercial databases and their extensions certifies more security in building such database systems. More standardization endeavors can be expected medium term, primarily for IBM’s and Oracle’s databases. Multimedia content management is the precondition for effective knowledge management of multimedia data - today’s IBM and Oracle multimedia implementations succeed in handling various company data and similar data available for in the Internet portal for example. Therefore IBM and Oracle multimedia implementations for images and texts are both recommended as a result of our work.

The following chapters will provide you an overview on relevant committees and standards, describe multimedia concepts, and summarize the theoretical and practical evaluation of multimedia capabilities. Finally, we will discuss challenges in multimedia databases, and propose a generic multimedia system architecture.

1. Introduction, scope & aim of document

This document highlights several aspects and recent developments on multimedia databases. Much attention is drawn on commercial software developments in this field. Multimedia databases have to store numeric, image, video, audio, text, graphical, temporal, relational and categorical data. We explore the support provided by existing object-relational systems. We focuses on the design and characteristics of such multimedia systems in terms of database support for multimedia applications and identify the limitations of existing state-of-the-art database management systems from the perspective of supporting multimedia applications. We first introduce the problem domain:

Multimedia databases and infrastructures gave gained tremendous attention in many application areas for the payoff in large interdisciplinary projects like medical information systems, geographic information systems, e-commerce or digital libraries. We will draw attention on special purpose database repositories within the DC corporate group with regard to data mining databases for discovering new knowledge.

The integral component for managing relational and object-oriented multimedia data is the database management system which provides mechanism to extract and represent the content of multimedia objects.

Most relational DBMS vendors have added capabilities to manipulate binary large objects (BLOBs) in the database. However, these implementations are not good enough for multimedia databases. The data is stored inside the BLOB as a non-interpreted byte stream. Because of this overly simple generalization, the DBMS does not have any knowledge concerning the content of the BLOB or its internal structure. Consequently, you cannot perform queries and operations on inherently rich and structured data types, such as images, video, web pages, hypertext, and word-processing documents. The operations and algorithms to manipulate these data types are not available to the query processing and indexing facilities of the DBMS. This means that the user application programs must perform the necessary processing on the contents of the BLOBs. Therefore content-based storage of multimedia gains more and more importance so that interesting database documents to a query can be found already on the database side. The second important requirement is a seamless integration of multimedia objects with traditional data in relational databases.

This document contains topics on multimedia object representation, content extraction and multimedia retrieval to cover all aspects of a complete multimedia database system. We will discuss how to represent structure and content of multimedia objects, how to extract meaningful features that capture the content of multimedia objects to build technically and semantically optimized indexing structures for content-based retrieval and to support data mining purposes. Another aspect is the development of user interfaces and query languages for extended multimedia data, e.g. editors with image input queries and the extension of standard SQL language for multimedia data [1][6], [7].

We consult several commercial databases for a comparative evaluation of this issues, namely IBM DB2 Database, Oracle 9i Database, Microsoft SQL-Server, Informix Database. Much attention is drawn on the evaluation of the two DaimlerChrysler’s strategic databases IBM DB2 Database and Oracle 9i Database. Microsoft SQL-Server and Informix are additionally evaluated where appropriate in order to give an overview of the complete spectrum of commercial multimedia extensions.

 

1. Overview

  Bibliography

1. Database Support for Multimedia Applications, Michael Ortega-Binderberger, Kaushik Chakrabarti, Sharad Mehrotra, in Image Databases, Search and Retrieval of Digital Imagery, edited by Vittorio Castelli and Lawrence D. Bergman (2003).
2. D.D. Chamberlin, A complete Guide to DB2 Universal Database. Morgan Kaufmann, Los Altos, California, (1998)
3. ORACLE, Oracle 8i Concepts, Release 8.1.6 Oracle (1999)
4. Informix, Getting Started with Informix Universal Server, Version 9.1, Informix (1997)
5. IBM Product Page, http://www.oreview.com/9704side.htm
6. Extending Rational DBMS, Martin Rennhackkamp (1998)
7. Digital Imagery: Fundamentals in Image Databases, Search and Retrieval of Digital Imagery edited by Vittorio Castelli and Lawrence D. Bergman (2003).
8. IBM Corporation. DB2 Universal Database. Image, Audio and Video Extenders –Administration and Programming, Version7 (2000)
9. Developing Java applications using DB2 Image and Audio Extenders, Ryan Cox, IBM Developer Solutions, http://www-106.ibm.com/developerworks/ibm/library/it/it-1001art29
10. IBM Software, Family Overview, DB2 Extenders

http://www-3.ibm.com/software/data/db2/extenders/index.html

11. ORACLE, Oracle Internet File System, User Guide, Release1.0, Oracle (2000)
12. Informix, Informix Universal Server, Datablade Programmer’s Manual Version 9.1 (1997)
13. ORACLE, Oracle Data Cartridges, Developer’s Guide, Release 8.1.5, Oracle (1999)
14. SQL Multimedia and Application Packages (SQl/MM), Jim Melton and Andrew Eisenberg (2001)
15. ISO/IEC 13249-1:2000,
Information technology, Database languages, SQL Multimedia and

Application Packages, Part 1: Framework, In-ternational Organization for

Standardization, (2000).

16. ISO/IEC 13249-2:2000, Information technology, Database languages, SQL Multimedia and

Application Packages, Part 2: Full-Text, In-ternational Organization for

Standardization, (2000).

17. ISO/IEC 13249-5:2001, Information technology, Database languages, SQL Multimedia and

Application Packages, Part 5: Still Image, In-ternational Organization for

Standardization, (2001).

18. 18. ISO/IEC 13249-6:2001, Information technology, Database languages, SQL Multimedia and

Application Packages, Part 6: Data Mining, Final Commitee Draft for Ballot, (2001)

19. SQL3 Object Model, Object Services and Consulting, Inc.,

http://www.objs.com/x3h7/sql3.htm (2002)

20. Informix, Media360, Any kind of content. Everywhere you need it. Informix (2000)
21. Informix product family, datablade modules,

http://www-3.ibm.com/software/data/informix/blades/, Informix (2003).

22. Informix product family, http://www-3.ibm.com/software/data/informix/ (2003).
23. Database Innovations, Laker Consulting Pty Ldt, http://www.laker.com.au/database.htm (2002).
24. Objektrelationales SQL, SQLJ & SQL/XML, Can Türker, (2003).
25. Object Database vs. Object-Relational Databases, Steve McClure, IDC Bulletin #14821E (1997).
26. About Databases Guide, http://www.databases.about.com/ (2003).
27. JDBC Data Access API, http://java.sun.com/products/jdbc/ (2003).
28. ANSI/ISO/IEC 9075-10:2000. ISO International Standard Database Language SQL-Part 10: Object Language Bindings (SQL/OLB), (2000).
29. ANSI/INCITS 331.1-1999. American National Standard for Information Technology: Database Language-SQLJ Part1: SQL Routines using the Java Programming Langage, (1999).
30. ANSI/INCITS 331.2-2000. American National Standard for Information Technology: Database Language-SQLJ Part2: SQL Types using the Java Programming Language, (2000).
31. SQLJ: The open sesame of Java database applications, Shirley Ann Stern, (1999).
32. Oracle 9i Database Release 2 Product Family , Oracle White Paper http://otn.oracle.com/products/oracle9i/pdf/9idb_rel2_prod_fam.pdf , (2003).
33. Oracle 8 Image Data Cartridges, Features Overview http://otn.oracle.com/products/oracle8/htdocs/ximg5fov.htm (1997).
34. Oracle8i interMedia Audio, Image, and Video User's Guide and Reference, Release 8.1.5

http://www.ucalgary.ca/it/oradoc/inter.815/a67299/toc.htm (1999).

35. Oracle Text: Features Overview http://technet.oracle.com/products/text/pdf/text_features_overview.pdf (2001).
36. Oracle Text An Oracle Technical White Paper, Oracle (2002).
37. Microsoft SQL Server, Product Information, Features by Edition

http://www.microsoft.com/sql/evaluation/features/choosing.asp (2001).

38. Roger Jennings, SQL Server 2000, Gains on Oracle, OakLeaf System (2002).
39. Rob Hawthorne, Oracle vs. SQL Server, Part 4 - The Softer Side of SQL Server 2000, (2002).

40. Simplifying Data Integration – Linking Java objects to relational databases,

Technology White Paper, Software Tree, Inc. (2003).

41. MSDM Microsoft Library, MS SQL Server, http://msdn.microsoft.com/library (2003).
42. James R. Hamilton and Tapas K. Nayak, Mircosoft SQL Server Full-Text Search,

IEEE (2001).

43. MSDM Microsoft Library, Transact-SQL Reference,

http://msdn.microsoft.com/library (2003).

44. Lynne Dunckley, Multimedia Databases, An Object-Relational Approach,

Addison-Wesley (2003).

45. Object Data Management Group, The Standard for Storing Object, http://www.odmg.org/, (2001).
46. ISO/IEC 11179, Specification, http://www.sdct.itl.nist.gov/~ftp/L8/11179/, (2000).
47. W3C Sematic Web Activity, (RDF) http:///www.w3.org/RDF/, (2003).
48. Ingo Schmitt, Retrieval in Multimedia-Datenbanksystemen, Institut für Technische und Betriebliche Informationssysteme, Otto-von-Guericke-Universität Magdeburg,

Draft to appear (2004).

49. MPEG ISO/IEC JTC1/SC29 WG11, http://ipsi.fraunhofer.de/delite/Projects/MPEG7/, (2003).
50. Common Warehouse Metamodel (CWM) Specification, Version 1.0, Part 1, (2001).
51. SQL Server 2000 vs. Oracle 9i, Alexander Chigrik, (2003).
52. Common Object Request Broker Architecture (CORBA), v3.0, OMG Syntax and Semantics Specification, (2002).
53. Oracle Technology Network, Index Types, http://otn.oracle.com/products/text/htdocs , (2003).
54. Oracle Technology Network, Daily F., http://otn.oracle.com/products/oracle9i/daily , (2002).
55. Comparison of DB2 Search Products,

http://www-3.ibm.com/software/data/db2/extenders/fulltextcomparison.html , (2003).

56. Susan Lawson, Object-Relational Programming with DB2, (2003).
57. University Information Technology Services, Database Administration Team, DB2 Standards Glossary, http://www.indiana.edu/~dbateam/databases/db2/standards/glossary.html , (2000).
58. Randall Davis, Howard Shrobe, Peter Szolovits, What is a Knowledge Representation,

MIT AI Lab (1993).

59. Ying Li, C.C. Jay Kuo, X. Wan, Introduction to Content-Based Image Retrieval, Overview of Key

techniques, (2002).

60. Christian Böhm, Stefan Berchtold, Hans-Peter Kriegel, Urs Michel, Multidimensional Index Structures in Relational Databases, Munich, Germany, (2000).
61. Vittorio Castelli, Multidimensional Indexing Structures for Content-Based Retrieval, IBM T.J. Watson Research Center, New York (2003).
62. Christos Faloutsos, Multimedia Indexing, Carnegie Mellon University, Pittsburgh, Pennsylvania, (2002).
63. Byoung-Kee Yi, H.V. Jagadish, and Christos Faloutsos. Efficient Retrieval of Similar Time Sequences under Time Warping. In IEEE Proc. of ICDE, Feb (1998).
64. Norman Walsh, What is XML?, http://www.xml.com/, (1998).
65. Josephine Cheng, Jane Xu, An end-to-end solution for storing and retrieving XML documents, IBM Silcon Valley, (2000).
66. Mark V. Scardina, XML Support in Oracle9i, An Oracle White Paper, Oracle Redwood Shores, (2000).
67. JavaSpaces Principles, Patterns and Practice, Official Java Sun Homepage, http://java.sun.com/docs/books/jini/javaspaces/ , (2003).
68. Eric Freeman, Susanne Hupfer, Ken Arnold, JavaSpaces Principles, Patterns and Practice, Addison Wesley, (1999).
69. Alejandro Jaimes, Shih-Fu Chang, A Conceptual Framework for Indexing Visual Information at Multiple Levels, DEE, Columbia University, New York, (2000).
70. S. Ravela, R. Manmatha, Retrieving Images by Appearance, University of Massachusetts, (1996).
71. Md. Farooque, Image Indexing and Retrieval, Documentation Research an Training Centre, Indian Statistical Institute, (2003).
72. Virginia E. Ogle, Michael Stonebraker, Chabot: Retrieval from a Relational Database of Images, University of California, Berkeley, (1995).
73. Informix Unleashed, Matt Bentley et al., Sams Publishing, http://docs.rinet.ru/InforSmes/fm/fm.htm
74. Fernando Pereira, The MPEG4 Standard: Evolution or Revolution , Instituto Superior Technico Lisboa, Portugal, (1996).
75. Ingrid Renz, Text Mining, The RIC/AM Technology Book, Ulm, (2002).
76. Paul Buitelaar, Daniel Sonntag, Knowledge Extraction, Hybrid approaches to data, text and image mining, in KEMIME FP6.2002.IST.1 IP project proposal, Germany, (2003).
77. Doreen Pittner, Analyse kommerzieller ORDB-Bild-Retrievalsysteme und Untersuchung notwendiger Erweiterungen zur Verbesserung der Retrieval-Qualität, Diploma thesis, University of Magdeburg / DC Research RIC/AM Ulm, (2004).
78. Frakes, W.B., & Baeza-Yates, R., Eds. (1992). Information Retrieval Data Structures & Algorithms. Prentice Hall, Englewood Cliffs, New Jersey.
79. TSpaces Intelligent Connectionware, IBM, http://www.almaden.ibm.com/cs/TSpaces/ (2003).
80. Alberto Bimbo, Pietro Pala, Visual Image Retrieval by Elastic Matching of User Sketches, IEEE Vol. 19, NO. 2, (1997).