INTERNATIONAL ORGANISATION FOR STANDARDISATION

ORGANISATION INTERNATIONALE DE NORMALISATION

ISO/IEC JTC1/SC29/WG11
CODING OF MOVING PICTURES AND AUDIO

ISO/IEC JTC1/SC29/WG11 N3751
La Baule, October 2000

Title: Introduction to MPEG-7
1
(v2.0)
Source: Requirements
Status: approved
Editors: Neil Day (Digital Garage Inc, JP), Jos M. Martnez (UPM-GTI, ES)


Introduction to MPEG-7

How many times have you seen science fiction movies such as 2001: A Space Odyssey and thought, Wow, were
so far away from having any of the fancy gadgets depicted in these movies! In 2001, Hal, the talking computer
intelligently navigates and retrieves information or runs complex operations instigated by spoken input. Or how
about using an image-based query, say an image of the motorbike used by Arnold Schwartzenegger in the movie T2,
to find images of similarly looking motorbikes. Dreams or reality?
As more and more audiovisual information becomes available from many sources around the world, many people
would like to use this information for various purposes. This challenging situation led to the need for a solution that
quickly and efficiently searches for and/or filters various types of multimedia material thats interesting to the user.
For example, finding information by rich-spoken queries, hand-drawn images, and humming improves the
user-friendliness of computer systems and finally addresses what most people have been expecting from computers.
For professionals, a new generation of applications will enable high-quality information search and retrieval. For
example, TV program producers can search with laser-like precision for occurrences of famous events or
references to certain people, stored in thousands of hours of audiovisual records, in order to collect material for a
program. This will reduce program production time and increase the quality of its content.
MPEG-7 is a multimedia content description standard, (to be defined by September 2001), that addresses how
humans expect to interact with computer systems, since it develops rich descriptions that reflect those expectations.
This document gives an introductory overview of the MPEG-7 standard. More information about MPEG-7 can be
found at the MPEG-7 website http://drogo.cselt.it/mpeg/ and the MPEG-7 Industry Focus Group website
http://www.mpeg-7.com. These web pages contain links to a wealth of information about MPEG, including many
publicly available documents, several lists of Frequently Asked Questions and links to other MPEG-7 web pages.


1
This document takes over the role played before by the MPEG-7: Context, Objectives and Technical Roadmap.
What Are the MPEG Standards?
The Moving Picture Coding Experts Group (MPEG) is a working group of the Geneva-based ISO/IEC standards
organization, (International Standards Organization/International Electro-technical Committee) in charge of the
development of international standards for compression, decompression, processing, and coded representation of
moving pictures, audio, and a combination of the two. MPEG-7 then is an ISO/IEC standard being developed by
MPEG, the committee that also developed the Emmy Award-winning standards known as MPEG-1 and MPEG-2,
and the 1999 MPEG-4 standard.
MPEG-1: For the storage and retrieval of moving pictures and audio on storage media.
MPEG-2: For digital television, its the timely response for the satellite broadcasting and cable television
industries in their transition from analog to digital formats.
MPEG-4: Codes content as objects and enables those objects to be manipulated individually or collectively on an
audiovisual scene.
MPEG-1, -2, and -4 make content available. MPEG-7 lets you to find the content you need.

Defining MPEG-7
MPEG-7 is a standard for describing features of multimedia content.
Qualifying MPEG-7:
MPEG-7 provides the worlds richest set of audio-visual descriptions
These descriptions are based on catalogue (e.g., title, creator, rights), semantic (e.g., the who, what, when, where
information about objects and events) and structural (e.g., the colour histogram - measurement of the amount of
colour associated with an image or the timbre of an recorded instrument) features of the AV content and leverages
on AV data representation defined by MPEG-1, 2 and 4.
Comprehensive Scope of Data Interoperability
MPEG-7 uses XML Schema as the language of choice for content description
MPEG-7 will be interoperable with other leading standards such as, SMPTE Metadata Dictionary, Dublin Core,
EBU P/Meta, and TV Anytime.

The Key Role of MPEG-7

MPEG-7, formally named Multimedia Content Description Inter-face, is the standard that describes multimedia
content so users can search, browse, and retrieve that content more efficiently and effectively than they could using
todays mainly text-based search engines. Its a standard for describing the features of multimedia content.
However
MPEG-7 will not standardize the (automatic) extraction of AV descriptions/features. Nor will it specify the search
engine (or any other program) that can make use of the description. It will be left to the creativity and innovation of
search engine companies, for example, to manipulate and massage the MPEG-7-described content into search
indices that can be used by their browser and retrieval tools.

Typical query examples enabled by MPEG-7 include:

Audio: I want to search for songs by humming or whistling a tune or, using an excerpt of Pavarottis voice, get a
list of Pavarottis records and video clips in which Pavarotti sings or simply makes an appearance.
Graphics: Sketch a few lines on a screen and get a set of images containing similar graphics, logos, and
ideograms.
Image: Define objects, including color patches or textures, and get examples from which you select items to
compose your image. Or check if your company logo was advertised on a TV channel as contracted.
Video: Allow mobile phone access to video clips of goals scored in a soccer game, or automatically search and
retrieve any unusual movements from surveillance videos.

MPEG-7 Technical Activities

It is important to note that MPEG-7 addresses many different applications in many different environments, which
means that it needs to provide a flexible and extensible framework for describing audio-visual data. Therefore,
MPEG-7 will define a multimedia library of methods and tools. It will standardize:
A set of descriptors: A descriptor (D) is a representation of a feature that defines the syntax and semantics of the
feature representation.
A set of description schemes: A description scheme (DS) specifies the structure and semantics of the relationships
between its components, which may be both descriptors and description schemes.
A language that specifies description schemes (and possibly descriptors), the Description Definition Language
(DDL): It also allows for the extension and modification of existing description schemes. MPEG-7 adopted XML
Schema Language as the MPEG-7 DDL. However, the DDL requires some specific extensions to XML Schema
Language to satisfy all the requirements of MPEG-7. These extensions are currently being discussed through
liaison activities between MPEG and W3C, the group standardizing XML.
One or more ways (textual, binary) to encode descriptions: A coded description is a description thats been
encoded to fulfill relevant requirements such as compression efficiency, error resilience, and random access.

Organization of MPEG-7 Descriptions

Over 100 MPEG-7 descriptions are currently being developed and refined. The relationships between the
descriptions in MPEG-7 are outlined in Figure 1.
The basic elements, at the lower level, deal with basic data types, mathematical structures, schema tools, linking
and media localization tools, as well as basic DSs, which are elementary components of more complex DSs. The
Schema tools section specifies elements for creating valid MPEG-7 schema instance documents and description
fragments.
In addition, this section specifies tools for managing and organizing the elements and datatypes of the schema.
Based on this lower level, content description and management elements can be defined. These elements describe
the content from several viewpoints. Currently five viewpoints are defined: creation and production, media, usage,
structural aspects, and conceptual aspects. The first three elements primarily address information thats related to
the management of the content (content management), whereas the last two are mainly devoted to the description
of perceivable information (content description).

Datatype &
structures
Link & media
localization
Basic DSs
Model
Basic elements
Navigation &
Access
Content management
Content description
Collection &
Classification
Summary
Variation
Content organization
Creation &
production
Media Usage
Conceptual
aspects
Structural
aspects
User
interaction
User
preferences
Schema
tools


Figure 1: Overview of MPEG-7 Multimedia Description Schemes (DSs)

Creation and Production: Contains meta information that describes the creation and production of the content;
typical features include title, creator, classification, and purpose of the creation. Most of the time this information is
author-generated since it cant be extracted from the content.
Usage: Contains meta information thats related to the usage of the content; typical features involve rights holders,
access rights, publication, and financial information. This information may be subject to
change during the lifetime of the AV content.
Media: Contains the description of the storage media; typical features include the storage format, the encoding of
the AV content, and elements for the identification of the media. Note: Several instances of storage media for the
same AV content can be described.
Structural aspects: Contains the description of the AV content from the viewpoint of its structure. The description
is structured around segments that represent physical, spatial, temporal, or spatio-temporal components of the AV
content. Each segment may be described by signal-based features (color, texture, shape, motion, audio) and some
elementary semantic information.
Conceptual Aspects: Contains a description of the AV content from the viewpoint of its conceptual notions.
The five sets of elements are presented here as separate entities, however, they are interrelated and may be partially
included in each other. For example, Media, Usage or Creation & Production elements can be attached to
individual segments involved in the structural description of the content.
Tools are also defined for navigation and access and there are another set of tools for Content organization which
addresses the organization of content by classification, by the definition of collections and by modeling. Finally,
the last set of tools is User Interaction which describes users preferences for the consumption of multimedia
content.

MPEG-7 Application Domains

The elements that MPEG-7 standardizes will support a broad a range of applications (for example, multimedia
digital libraries, broadcast media selection, multimedia editing, home entertainment devices, etc.). MPEG-7 will
also make the web as searchable for multimedia content as it is searchable for text today. This would apply
especially to large content archives, which are being made accessible to the public, as well as to multimedia
catalogues enabling people to identify content for purchase. The information used for content retrieval may also be
used by agents, for the selection and filtering of broadcasted "push" material or for personalized advertising.
Additionally, MPEG-7 descriptions will allow fast and cost-effective usage of the underlying data, by enabling
semi-automatic multimedia presentation and editing. All domains making use of multimedia will benefit from
MPEG-7 including,

Digital libraries, Education (image catalogue, musical dictionary, Bio-medical imaging catalogues)
Multimedia editing (personalised electronic news service, media authoring)
Cultural services (history museums, art galleries, etc.),
Multimedia directory services (e.g. yellow pages, Tourist information, Geographical information systems)
Broadcast media selection (radio channel, TV channel,)
Journalism (e.g. searching speeches of a certain politician using his name, his voice or his face),
E-Commerce (personalised advertising, on-line catalogues, directories of e-shops,)
Surveillance (traffic control, surface transportation, non-destructive testing in hostile environments, etc.),
Investigation services (human characteristics recognition, forensics),
Home Entertainment (systems for the management of personal multimedia collections, including
manipulation of content, e.g. home video editing, searching a game, karaoke,)
Social (e.g. dating services),

Typical applications enabled by MPEG-7 technology include:

Play a few notes on a keyboard and retrieve a list of musical pieces similar to the required tune, or
images matching the notes in a certain way, e.g. in terms of emotions.
Draw a few lines on a screen and find a set of images containing similar graphics, logos, ideograms,...
Define objects, including colour patches or textures and retrieve examples among which you select the
interesting objects to compose your design.
On a given set of multimedia objects, describe movements and relations between objects and so search
for animations fulfilling the described temporal and spatial relations.
Describe actions and get a list of scenarios containing such actions.
Using an excerpt of Pavarottis voice, obtaining a list of Pavarottis records, video clips where Pavarotti
is singing and photographic material portraying Pavarotti.

Examples of MPEG-7 Applications
The following three applications are examples of the type of solutions that MPEG-7 can solve, (references at the
end of this document). For purposes of simplification, these applications are categorized as Visual, Audio and
Multimedia applications although it is recognized that there are cases where functions between these categories
overlap with each other. Since the development effort of MPEG-7, however, highlights audio, visual, and
multimedia description schemes and descriptors, it is suffice for the moment to use these categories. These
application examples represent development work in progress and are based on actual demos that are being
constructed. There are many more applications being developed around the world (as evidenced by the first
MPEG-7 Awareness Event, October 2000: www.mpeg-7.com) and the reader is invited to contact MPEG-7
representatives for more information (see reference at the end of this document).

I. Visual
Query by Image

Figure 2 shows possible ways to search for visual content using the inherent structural features of an image. In this
example there are four image features detailed. The color histogram feature (1) of an image allows me to search for
images that have the same color. Note, the position of the colors is not important but rather the amount of similar
color in the image is important. The next feature, spatial color distribution (2) allows me to search for images where
the location of the same color is important. You can see that the added object in the right-bottom flag does not affect
this type of search. You can additionally search for images that have a similar edge or contour profile as in the
spatial edge distribution (3) search technique. Note, color does not make a difference to this type of search. Finally,
you can see an example of searching by object shape (4). Here, the color and edge profile are not important.
3. Spatial edge
distribution
2. Spatial color
distribution
4. Object shape
1. Color histogram

Figure 2: Search using image features

II. Audio
Music Retrieval By Melodic Query

Identifying a musical work from a melodic fragment is a task that most people are able to accomplish with relative
ease. For some time now researchers have worked to give computers this ability as well, which has come to be
known as the "query-by-humming" problem. To accomplish this, it is reasonable to study how humans are able to
perform this task, and to assess what features we use to determine melodic similarity. Research has shown that
melodic contour is an important feature in determining melodic similarity, but it is also clear that rhythmic
information is important as well.








Beat
Contour: 2 -1 -1 -1 -1 -1 1
Number: 1 4 5 7 8 9 9 10



Figure 3: Audio features used in Search by Humming

The application being developed using the MPEG-7 description scheme for melody incorporates melodic contour
and rhythmic information as the primary representation for music search and retrieval.
Additional front-end processing (to process queries), a medium-sized database of music, and a search engine (for
finding appropriate matches) have also been implemented to complete the full query-by-humming system.

III Multimedia

Wireless Images Retrieval using Speech Dialogue Agent

One typical usage scenario for this application is when a user wishes to know traffic information at certain locations
to help him/her navigate their travel route, (see figure 4). Here an agent in the client terminal (containing specific
sentence utterances in its dictionary) recognizes the user's utterance in English/Japanese and sends a query profile
to the server using a wireless transceiver channel (32kbps). The server will retrieve the requested images and
deliver the compressed video bitstream (H.263) to the client. Then the client agent will reply with a synthesized
voice and display the images. Currently, the original format of the metadata is being used, but the MPEG-7 format
would be used in the near future for all the clients, servers, and channels.


This is the current status
























Figure 4: Wireless Images Retrieval using Speech Dialogue Agent

What is the status of MPEG-7?

Currently MPEG-7 concentrates on the specification of description tools (Descriptors and Description Schemes),
together with the development of the MPEG-7 reference software, known as XM (eXperimentation Model). The
XML Schema Language was chosen as the base for the Description Definition Language (DDL).
The MPEG-7 Audio group develops a range of Description Tools, from generic audio descriptors (e.g., waveform
and spectrum envelopes, fundamental frequency) to more sophisticated description tools like Spoken Content and
Timbre. Generic Audio Description tools will allow the search for similar voices, by searching similar envelopes
and fundamental frequencies of a voice sample against a database of voices. The Spoken Content Description
Scheme (DS) is designed to represent the output of a great number of state of the art Automatic Speech Recognition
systems, containing both words and phonemes representations and transition likelihood. This alleviates the
problem of out-of-vocabulary words, allowing retrieval even when the original word was wrongly decoded. The
Timbre descriptors (Ds) describe the perceptual features of instrument sound, that make two sounds having the
same pitch and loudness appear different to the human ear. These descriptors allow searching for melodies
independently of the instruments.
The MPEG-7 Visual group is developing four groups of description tools: Color, Texture, Shape and Motion. Color
and Texture Description Tools will allow the search and filtering of visual content (images, graphics, video) by
dominant color or textures in some (arbitrarily shaped) regions or the whole image. Shape Description Tools will
facilitate query by sketch or by contour similarity in image databases, or, for example, searching trademarks in
registration databases. Motion Description Tools will allow searching of videos with similar motion patterns that
can be applicable to news (e.g. similar movements in a soccer or football game) or to surveillance applications (e.g.,
detect intrusion as a movement towards the safe zone).
The Multimedia Description Schemes group is developing the description tools dealing with generic and
audiovisual and archival features. Its central tools deal with content management and content description. Content
Management description tools cover the viewpoints of Media, Creation and Production, and Usage. Media
description tools allows searching for preferred storage formats, compression qualities, and aspect ratios among
others. Creation and Production descriptions tools cover the typical archival and credits information (e.g., title,
creators, classification). Usage description tools deal with description related to the use of the described content
(e.g. rights, broadcasting dates and places, availability, audience, financial data). The Content Description ones
cover both structural and conceptual viewpoints. Structural description tools provide segmentation, both spatial
and temporal, of the content. This allows, among other functionalities, assigning descriptions to different regions
and segments (e.g., to provide the means for a segment annotation instead of only a global one) and providing
importance rating of temporal segments and regions (e.g., allowing to differentiate among regions of the content
for adaptive coding with different quality). Conceptual description tools allow providing semantic based
description (e.g., linguistic annotations of the content, and object and event description from a knowledge
viewpoint). Besides the Content Description and Content Management description tools, there are others targeted
to content organization (e.g., to organize an archive of images descriptions in a repository), navigation and access
(e.g., to display a summary of videos through relevant short sequences or keyframes for quick browsing), and user
preferences (e.g. for agent based selection or filtering of favorite programs).

MPEG-7 in the 21
st
century Media landscape: the Uniqueness of MPEG-7

MPEG-7 is about the future of media in the 21
st
century. This is not an overstatement. MPEG-7 provides a
comprehensive and flexible framework for describing the content of multimedia. To describe content implies
knowledge of elements it consists of, as well as, knowledge of interrelations between those elements. The most
straightforward application is multimedia management, where such knowledge is prerequisite for efficiency and
accuracy. However, there are other serious implications. Knowledge of the structural features of multimedia
information as well as its semantic features will help generate solutions that will provide more comprehensive and
accurate indexing and search applications, (leading to greater ability for content manipulation, content reuse - and
thus new content creation). Many issues, it is true, remain including copyrights issues and interoperability between
applications and systems that wish to adhere to the MPEG-7 standard. But such issues are balanced by incredible
economical, educational, and ergonomic benefits that will be brought by MPEG-7 technology. Potential concerns
will be resolved, and after some years hence, we will not be able to imagine media without MPEG-7 technologies.

REFERENCES

MPEG-7 Web Sites
There are a number of documents available at the MPEG (http://drogo.cselt.it/mpeg/) Home Page. Information
more focused to industry is also available at the MPEG-7 Industry Focus Group, (http://www.mpeg-7.com) Web
site.

MPEG-7 Personal Contacts
For Technical Issues:
Requirements: Rob Koenen ([email protected])
Audio: Adam Lindsay ([email protected])
Visual: Thomas Sikora ([email protected])
Multimedia DS: Philippe Salembier ([email protected])
Systems: Olivier Avaro ([email protected])
XM Software Implementation: Stephan Herrmann ([email protected])

MPEG-7 Industry Focus Group: Neil Day ([email protected])
Editors: Neil Day, ([email protected]),
Jos M. Martnez ([email protected])

Highlighted MPEG-7 Applications

Audio: Query by Humming
Contact: Youngmoo Kim, Email: [email protected], http://sound.media.mit.edu/~moo
Machine Listening Group, MIT Media Lab., Boston, USA.

Visual: Query by Image
Contact: Soo-Jun Park, Email: [email protected], http://sir.etri.re.kr/~soop
ETRI-CSTL, 161 Kajong-dong, Yoosung, Taejon, 305-350, Korea

Multimedia: Wireless Images Retrieval using Speech Dialogue Agent
Contact: Mikio Sasaki: Email: [email protected]
DENSO CORPORATION, 500-1 Minamiyama, Komenoki-cho, Nisshin-shi, Aich-ken,470-0111, Japan.