Paper on Multimedia Database Management System

Contents 1. Introduction 2. Purpose of Multimedia Database management system 3. Requirements for the multimedia D M! ". Multimedia #b$ects % Characteristics ".1 &ccess Dimensions of the Media #b$ects '. Multimedia Database Management !ystem % Components Physical storage (ie) Conceptual Data (ie)

Distributed (ie) *iltered (ie) +sers, (ie) -. Conclusions .. Reference

1. Introduction
A multimedia database management system (MMDBMS) is a framework that manages different types of data potentially represented in a wide diversity of formats on a wide array of media sources. he multimedia DBMS provides support for multimedia data types! plus facilities for the creation! storage! access! "uery and control of the multimedia database.

2. Purpose of Multimedia Data !ystem

ase Management

A Multimedia database management system provides a suitable environment for using and managing multimedia database information. #t must support the various multimedia data types ! in addition to providing facilities for traditional DBMS functions like database definition and creation! and retrieval! data access and organi$ation! data independence! privacy! integration! integrity control! version control and concurrency support. %sing the general functions provided by a traditional DBMS as a guide! we can describe the purposes of a multimedia DBMS as follow Integration % &nsures that data items needs not be duplicated during different program invocation re"uiring the data. Data Independence% Separation of the database and the management functions from the application programs. Concurrency Control% &nsures multimedia database consistency through rules! which usually impose some form of e'ecution order on concurrent transactions. Persistence% he ability of data ob(ects to persists (survive) through different transactions and program invocations. Pri/acy% )estrict unauthori$ed access and modification of stored data. Integrity control% &nsures consistency of the database state from one transaction to another through constraints imposed on transaction. Reco/ery% Methods needed to ensure that results of transaction that fail do not affect the persistent data storage.

0uery !upport% &nsure that that the "uery mechanisms are suited for multimedia data. (ersion Control% *rgani$ation and management of different version of persistent ob(ects! which might be re"uired by applications. #n concurrency control! a transaction is a se"uence of instructions e'ecuted either completely or not at all. #n the latter case! the database is restored to its previous state. Defining the appropriate granularity for concurrency is a problem in multimedia database. raditional database use a single record or table as the unit of concurrency. Multimedia databases typically use a single ob(ect as the logical unit of access. hus the single multimedia ob(ect could form the unit of concurrency. #n achieving persistence! a simple method is to store the multimedia files in some operating system files. +owever! the huge data volumes make this approach costly to implement. Moreover the system also needs to store the multimedia metadata and possibly composite multimedia ob(ects. hus most multimedia DBMS classify the data as either persistent or transient and store only persistent data after transaction updates. ,ersistent data after transaction updates transient data are used only during program or transaction e'ecution and are removed afterwards. raditionally! a "uery selects a subset to the data ob(ects of the data ob(ect based on the user-s description of what data to access. A "uery usually involves various attributes. ,ossibly 1ey)ord2based or Content oriented and is usually interactive. hus functions for relevance feedback and "uery formulation. Similarity matches and mechanisms for displaying ranked results are important in multimedia DBMS .ersion control becomes important when a persistent multimedia ob(ect is updated or modified as some application might need to access previous states of the ob(ect. A DBMS provides such access through versions of the persistent ob(ects. /or a multimedia DBMS the huge volumes of data reinforces the importance of efficiently organi$ing such versions. Moreover the available storage might limit the provision of version in addition! version management may

involve not only version of single ob(ects but also versions of the comple' ob(ects that make up the multimedia database. he special nature of multimedia data also makes it important to support new special functions. hese include ob(ect composition and decomposition! management of huge volumes of multimedia data! effective storage management and information retrieval and handling of spatial and temporal data ob(ects

3. Requirements for the multimedia D M! /or multimedia DBMS to serve its e'pected purpose! it must meet certain special re"uirements. raditional DBMS capabilities +uge capacity storage management #nformation retrieval capabilities Media #ntegration! 0omposition presentation Multimedia 1uery support Multimedia interface and interactivity ,erformance

and

". Multimedia #b$ects% Characteristics

Multimedia databases can be accessed by "ueries on any of the ob(ects composing the databases. he properties of these media ob(ects distinguish the needs of a multimedia database management system from those of a traditional one as below2 3a4 5e6t Data e't data is often represented as strings. +owever! te't as used in multimedia document systems! includes structural information such as title! author etc. +ence one needs a language environment to reflect the structural composition of the te't data. Standard 3enerali$ed Markup 4anguage (S3M4) is a document representation language defined by #nternational Standards *rgani$ation (#S*). Another environment named the +ypermedia5 ime6based Structuring 4anguage (+ytime) has also been defined to include support for hypermedia documents (hyperte't with multimedia ob(ects)! with links and support for inclusion of multimedia ob(ects in a te't document specifications. S3M4 together with +y ime can be used for developing multimedia document.

3b4 &udio Data Audio data has an inherent time dependency associated with it. he time scale associated with audio ob(ects has to be uniform for a meaningful interpretation. Audio has to be digiti$ed before it can be processed. Si$e of digiti$ed audio depends on the techni"ue used! which in turn depends on the desired audio "uality. /or instance! a normal voice "uality digiti$ation is done at 7 k+$ with 7 bits per sample! and hence it produces 89:b5s of data. 0D "uality digiti$ation is carried out at 99.; :+$ sampling rate with ;8 bits per sample and hence produces

;.9Mb5s. Digiti$ed audio can be effectively compressed to reduce storage re"uirements. 3c4 Image Data #mage data represents digiti$ed drawings! paintings! or photographs. he si$e of a digiti$ed image depends on the re"uired "uality. 0olor images and photographs re"uire more storage space. ypically! a color image or a photograph needs the )3B components of each pi'el to be stored. Depending on the color scale chosen! one might need 7 bits per color component implying <9 bits per pi'el. +ence ;=<9>;=<9 pi'el image! a storage space of <9 Mbits is needed. 0ompression schemes are used to reduce the volume of data that needs to be stored. Different compression algorithms as well as storage representations can be employed and this results in different formats of digiti$ed images and photographs. ?oint ,hotographers &ncoding 3roup (?,&3) is one such format for images! which has been standardi$ed by #S*. *ther popular formats include 3raphic #nterchange /ormat (3#/) and ag #mage /ormat ( #//).

3d4 .ideo Data .ideo data represents the time dependent se"uencing of digiti$ed pictures or images called video frames. he number of video frames for one second depends on the standard that is employed. @ S0 (@ational elevision System 0ommittee) standard employs A= frames5second while ,A4 (,hase Alternation 4ine) employs <B frames5second. /urther! the pi'el si$e of a frame depends on the desired "uality. @ormal @ S0 frames are B;< > 97= pi'els in si$e! +D . (+igh Definition elevision) frames employ ;=<9 > ;=<9 pi'els. he number of bits needed per pi'el reflects the "uality of digiti$ed video frame. Digiti$ed video re"uires large storage space. 0ompression schemes need to be employed to reduce the volume of data to be stored.

Motion ,icture &ncoding 3roup (M,&3) is a standard that has been specified by the #S* for compression and storage of video. he standard M,&3 < specifies the methodology for storing audio along with compressed video.

".1 &ccess Dimensions of the Media #b$ects Cith reference to the process of accessing the contents! media ob(ects can be considered as one of the following2

(a) 1-Dimensional Objects e't and audio have to be accessed in a contiguous manner (as AS0## strings or signal waves). +ence! te't and speech can be considered as ;6dimensional ob(ects.

(b) 2-Dimensional Objects Access to image data can be done with reference to the spatial locations of ob(ects. /or e'ample! a "uery can search for an ob(ect that is to the right of or below a specified ob(ect. So image ob(ects can be considered as <6dimensional since it has a spatial content. (c) 3-Dimensional Objects .ideo has spatial characteristics as well as temporal characteristics. Access to video can be done by describing the temporal as well as the spatial content. /or e'ample! a "uery can ask for a movie to be shown from ;= minutes after its starting point. +ence! video can be considered as a A6 dimensional ob(ect. he access dimension of an ob(ect! in a way! describes the comple'ity in the process of searching. /or ;6dimensional ob(ects! such as te't and audio! the access is limited to the keywords that appear as part of te't or speech.

/or image! the access is done by specifying the contents as well as their spatial organi$ation. #n a similar manner! access to video should comprise of the se"uencing of video frames in the time domain.

'. Multimedia Database Components

Management

!ystem%

/igure shows the components of a multimedia database management system. he physical storage view describes how multimedia ob(ects are stored in a file system. Since multimedia ob(ects are typically huge! we need different techni"ues for their storage and retrieval. his view deals with the issue of providing fast access to stored data by means of inde' mechanisms. Multimedia ob(ects can be stored in different systems and users might access stored data over computer networks. his leads to a distributed view of multimedia databases. %sers can "uery multimedia databases in different ways! depending on the type of information they need. hese "ueries provide a filtered view of the multimedia databases to the users by retrieving only the re"uired ob(ects. he ob(ects retrieved from the databases have to be appropriately presented! providing the user-s view of the multimedia database. hough these views are true for a traditional database management system! the diverse characteristics of media ob(ects introduce many interesting issues in the design of a multimedia database management system.

*ig. Components In/ol/ed In Multimedia Database

Physical !torage (ie) he main issue in the physical storage of multimedia ob(ects is their si$es. Size of objects influences both the storage capacity re"uirements and the retrieval bandwidth (in terms of bps) re"uirements. he table below describes the si$e and the retrieval disk bandwidth re"uirements for different media! based on their

format of representation. he disk bandwidth re"uirements of discrete media such as te't and images depend on a multimedia database application. his is because these media do not have any inherent temporal requirements. he bandwidth re"uirements of discrete media might depend on the number of images or the number of pages of te't! that an application needs to present within a specified interval of time. +owever! continuous media such as video and audio have inherent temporal requirement! eg. A= frames5s for @ S0 video. hese temporal re"uirements imply that an uncompressed B minutes video clip ob(ect will re"uire A== times its storage space for ; second.

he disk bandwidth re"uirements (for storage and retrieval) in the case of continuous media are proportional to their temporal re"uirements! since the temporal characteristics dictate the storage as well as the presentation of the data. /urther! the stored video data might be accessed by multiple users simultaneously. +ence! these characteristics of video demand new capabilities from the file system and the operating system.

*ile !ystem Requirements A file system for multimedia data storage should have the following capabilities for (a) +andling huge files (of the order of 3igabytes) (b) Supporting simultaneous access to multiple files by multiple users. (c) Supporting the re"uired disk bandwidth. he caching strategies followed by a file system should also support these re"uirements. he file system can provide new application programming interfaces apart from the traditional ones such as open! read! write! close and delete. he new application programming interfaces can support play! fast forward! and reverse for continuous media such as video. #perating !ystem Requirement *perating system supporting multimedia applications should have capabilities for handling real time characteristics for (a) Scheduling of application processes (b) 0ommunication between an application process and the operating system kernel. /or real time scheduling! the operating system might have to reserve the resources re"uired for an application process. his implies that depending on the availability of resources! an

application process may or may not be admitted for e'ecution by the operating system. Another important re"uired feature is the reduced overhead in the communication between application processes and the operating system kernel. his overhead directly affects the performance of applications.

Conceptual Data (ie) ,hysical storage of multimedia ob(ects deals with raw digiti$ed data. #n this stage! multimedia ob(ects are in binary form. hese ob(ects are ac"uired (from devices) and created (digiti$ed! compressed and stored) independent of its contents. /or using these ob(ects as meaningful data! one needs to identify their content. he description of the ob(ects- content! called metadata is sub(ective in nature and is dependent on the media type as well as the role of an application. /or instance! the description of the person-s nose is sub(ective. he description also depends on the role of the application. /eature description of a facial image may not be needed for a particular application and hence the database may not carry such description. Meaningful description of video clips has to be identified and stored in the database. /or e'ample! consider a video clip of a movie. he se"uence of frames contains actors! actresses! the background of the scene! action going on in the scene etc. +ence a description of the video clip might contain the description of the characters acting in the movie! the background and the action part of it. he figure below shows this e'ample.

*ig% 76ample Description #f a (ideo Clip he conceptual data view of raw multimedia data helps in building a set of abstraction. hese abstractions form a data model for a particular application domain. /or fast accesses! we need indexing mechanisms to sort the data according to the features that are modeled. A multimedia database may be composed of multiple media ob(ects whose presentation to the user has to be properly synchroni$ed. hese synchronization characteristics are described by temporal models. +ence the conceptual view of multimedia data consists of the following components2 (a) (b) (c) (d) (e) Metadata #nde'ing mechanisms emporal models Spatial models Data models

Metadata (a) Metadata deals with the content! structures! and semantics of media ob(ects. (b) he creation of metadata depends on the media type and the type of information which an application wants to describe as part of the metadata. (c) /rom the maintenance of multimedia database point of view! it is important that techni"ues for automatic (or semi6 automatic) generation of metadata for each media type are available.

(d) /or video media! the techni"ues should identify camera shots! characters in a shot! background of a shot etc. (e) /or image data! techni"ues should e'tract and describe the features of interest. #n a similar manner! recognition techni"ues might be needed for identifying keywords in audio and te't data. Inde6ing Mechanisms (a) Multimedia databases need inde'ing mechanisms to provide fast access. he techni"ues developed for traditional databases do not serve this purpose fully! since new ob(ect types have to be dealt with. (b) he inde'ing mechanism should be able to handle different features of ob(ects such as color or te'ture. 5emporal Models (a) #t describes the time and duration of presentation of each media ob(ect as well as their temporal relationships to other media ob(ects. (b) /or instance! /igure ;.< describes the temporal relationship among the ob(ects composing of .oD database. +ere as an e'ample! the video ob(ect D; has to be presented at time t; for a duration tA6t; and has to be synchroni$ed with the presentation of audio ob(ect E;. !patial Models (a) #t represents the way media ob(ects are presented by specifying the layout of windows on a monitor. (b) /igure ;.8 shows a possible organi$ation of windows for presenting the ob(ects in the .oD database.

Data Models (a) *b(ect oriented approach is normally used to represent the characteristics of ob(ects! metadata associated with them! their temporal and spatial re"uirements.

Distributed (ie) 4ike other information! multimedia data can also be distributed over computer networks. +uge si$es of media ob(ects re"uire large bandwidths or throughput (bps). )eal time nature of the ob(ects needs guarantees on end6to6end delay and delay (itter. &nd to end delay specifies the ma'imum delay that can be suffered by data during communication. Delay (itter describes the variations in the end to end delay suffered by the data. 3uarantees on end6to6end delay and delay (itter are re"uired for smooth presentation of continuous media ob(ects such as audio and video. &'isting communication protocols address the needs of more traditional applications such as file transfer! remote login! and electronic mail. hese applications do not have any real time re"uirements and so there is little need for large bandwidth. So distributed multimedia applications re"uire new generation of protocols. Due to their huge si$es! not many media ob(ects can be buffered by the client. Also! the bandwidth offered by the network is not unlimited.

+ence! based on the temporal relationships! the buffers re"uired and the available network bandwidth! the client needs to identify a retrieval schedule for re"uesting ob(ects from the server.

*iltered (ie) /iltered view of a multimedia database is provided by a user-s "uery to get re"uirement information. he "uery can be on any of the media that compose a database. %ser-s "uery can be of the following types2 (a) (b) (c) (d) (e) 1uery on the content of media ob(ects 1uery by &'ample ime inde'ed "ueries Spatial "ueries Application specific "ueries

Content ased 0uery 1ueries on the content of media ob(ects typically re"uire a search on the metadata associated with the ob(ects. 1ueries ; and < of Section ;.<.; belong to this category.

0uery by 76ample (a) 0onsidering the .oD server application! users can make "ueries by e'ample such as - 3et me the movie in which this scene (an image) appears - 3et the movie where this video clip occurs - Show me the movie which contains this song (b) his refers to the multimedia ob(ect that is used as an e'ample. he multimedia database management system has to process the e'ample data (this ob(ect) and find one that matches it i.e. the input "uery is an ob(ect itself. (c) he re"uirement for similarity can be on different characteristics associated with the media ob(ect. (d) he re"uired similarity matching between the "ueried ob(ect and database ob(ects can be e'act or partial.

(e) #n the case of partial matching! we need to know the degree of mismatch that can be allowed between the e'ample ob(ects and the ones in the database. 5ime inde6ed 0ueries (a) Since multimedia database is composed of time6dependent or continuous media! users can give "ueries in the temporal dimension as well. (b) /or e'ample! a time inde'ed "uery for a .oD server can be 2 Show me the movie 30 minutes after its start. !patial 0ueries (a) Media ob(ects such as image and video have spatial characteristics associated with them. (b) +ence! users can issue a "uery like the following one2 Show me the image where Billgets is seen to the left of Ajim premji

&pplication !pecific 0ueries (a) Multimedia database are highly application specific. 1ueries therefore can be application specific too. (b) Ce can consider database such as medical or geographic information database. (c) %sers can ask "ueries such as Show me the video where the tissue evolves into a cancerous one

+sers, (ie) %ser-s view of a multimedia database management system is characteri$ed by the following re"uirements2 (a) %ser "uery interface (b) ,resentation of multimedia data (c) %ser interaction during presentation

+ser 0uery Interface (a) 1uery interface should allow users to "uery by content! e'ample! time! space! or a combination of these possibilities. (b) /or "ueries by e'ample! the user "uery interface has to obtain the e'ample ob(ect from appropriate devices eg image from scanner. (c) #n case of partial matching of the resolved "ueries! the "uery interface can suggest ways to modify the "uery to get e'act matches.

Presentation of Multimedia Data (a) Media ob(ects can be of different format. /or e'ample! images can be stored in tiff or gif format. (b) #n some cases! there might be a necessity to convert data from one format to another format before presentation. (c) Also multimedia ob(ects composing the responses to user-s "uery have associated temporal constraints. (d) hese constraints specify the time instants and the durations of the presentations of various multimedia ob(ects. (e) #n a similar manner! the presentation of multimedia ob(ects may have spatial constraints. (f) hese constraints describe the layout of windows on the user-s screen for presentation of different ob(ects.

+ser Interaction During Presentation (a) %ser can interact during the presentation of multimedia ob(ects. he interaction is more comple' (compared to traditional databases) since multiple media ob(ects are involved. (b) /or instance! devices such as microphone and video camera can be used for speech and gesture recognition. (c) +ence! simultaneous control of different devices and handling of user input is re"uired. he input from the user can be of following types2 6 modify the "uality of the presentation! reduction or magnification of the image 6 direct the presentation! skip! restart etc.

&pplication %

#n 3eneral! Multimedia database management system application can be found wherever there is a need to manage multimedia data cost effectively. hus multimedia DBMS have found applications in such diverse areas as education (digital libraries! training! presentation! distance learning )! healthcare( telemedicine! health information management! medical image system) entertainment ( video6on6demand! music databases ! interactive .). #nformation dissemination (news6on6demand! advertising! . broadcasting)!and manufacturing ( distributed manufacturing! distributed collaborative authoring). *ther areas may include finance! video conferencing! electronic publishing! electronic commerce! and graphics information system. Most of multimedia DBMS already e'ist. Most are e'tension from e'isting ob(ect oriented or relational DBMS.

-. Conclusions /rom this we can conclude that the database management for multimedia is re"uired and it can be managed like normal data that will very helpful in the current era of multimedia application. @umber of multimedia database system already e'ists with some limitation but we can e'pect new system with better capabilities in the near future. /uture trend include performing inde'ing ! retrieval and browsing directly on the compressed data ! especially for video dataF video data managementF multimedia "uery language F uniform inde'ing frame work for the different data typesF content base image and video retrievalF and multimedia transport and deliver overt the internet.

.. Reference
http255www.hitachi.com http255foe.mmu.edu.my www.ssgr.it www.fsksm.utm.my