MPEG-7
Multimedia Content Description Interface
by
Jozsef Vass
2/2/98
Table of Contents
Overview
- Introduction
- History
- MPEG-7 Documents
- Schedule
- MPEG-7 Objectives
- Scope of MPEG-7
- MPEG-7 Requirements
- MPEG-7 Applications
- Possible Research Directions
- Case Studies -- QBIC, Columbia University
Introduction
- Not a compression standard
- MPEG-7 will be associated with the indexing of the available audio-visual
(AV) information
- Before usage, information needs to be located -> quickly
and efficiently search for various types of multimedia information
- Search for textual information exists (WWW search engines) but locating
multimedia material is very difficult
- Information about content -> metadata
- Why MPEG-7? -> MPEG-5 would continue the series and MPEG-8
would give a binary pattern
History
| Meeting | Date | Event
|
| San Jose | February 1998 |
|
| Fribourg | October 1997 | Refinements
|
| Stockholm | July 1997 | Documents
|
| Bristol | April 1997 | MPEG-7 Seminar
|
MPEG-7 Documents
- MPEG-7 Requirements Group, "MPEG-7": Context and objectives"
- MPEG-7 Requirements Group, "Second/third draft MPEG-7 requirements"
- MPEG-7 Requirements Group, "MPEG-7" applications document."
Schedule
| Call for Proposals | November 1998
|
| Working Draft | July 1999
|
| Committee Draft | March 2000
|
| Draft International Standard | July 2000
|
| International Standard | November 2000
|
MPEG-7 Objectives
- Allow fast and efficient searching for multimedia material of user's interest
- Specify a standard set of description schemes and descriptors
to describe various types of
multimedia information
- Specify encoding of description schemes and descriptors
- Descriptors are associated with the content itself
- Descriptors must be meaningful in the context of application
- MPEG-7 data can be located anywhere
- Descriptors are independent of the encoding of the content
- Any type of AV material can be retrieved by any type of query material
Scope of MPEG-7

Scope of MPEG-7
- MPEG-7 is general
- Only the descriptor is standardized
MPEG-7 Requirements
Requirements are divided into 1) common audio and visual requirements,
2) visual requirements, and 3) audio requirements
- Query Classes: MPEG-7 shall support audio and visual descriptors
that allows different types of queries:
-- Textual Description Classes: keywords, title, author, structure, etc.
-- Visual Description Classes: color, texture, shape, volume, etc.
-- Audio Description Classes: frequency contour, timbre, harmony, textual
content, etc.
- Content-Based Retrieval: Support effective and efficient retrieval
of multimedia based on their content
- Types of Features: Various types of features ranging from low level
to high level features
- Feature Hierarchy and Scalability: Coarse-to-fine refinement of features
- Distributed Databases: Support of simultaneous and transparent retrieval
- Robustness to Information Errors and Loss
- Ability to export existing databases into MPEG-7 format
- Copyright Information
MPEG-7 Applications
Application domains:
- Education
- Journalism
- Entertainment
- Tourism
- Medical Applications
- Archives (Film, Video, and Audio)
- Shopping
- Social
Possible applications:
- Intelligent Multimedia Presentation: By using knowledge of context,
application, location of user the system can adapt to the user
(similar to Web personalizing)
- Storage and Retrieval of Video Databases: Enormous interest to
search end retrieve from existing huge radio and TV archives
- Teleshopping: Humans remember better to visual information than textual.
Internet catalogs could be searched much easier by using visual query.
- Entertainment: For example Karaoke user interface can be made more
convenient by query by example (sing part of song)
- Medical: Physicians can search huge medical archives for
similar symptoms by query by example
Research Directions
- Combination of low level information (extracted automatically)
with high level description (mostly obtained manually)
- Data structures
- How MPEG-4's AV objects can be extended to support the requirements of
MPEG-7?
- Language independence
Case Study - QBIC
- QBIC: Query by Image Content
- Computing image features and retrieval based on these features
- Can be extended to video retrieval by using key frames or mosaics
- More information: http://wwwqbic.almaden.ibm.com/
Case Study - Finding Images/Video Archives
- Image-Based Visual Query Project
- WWW-based
- Retrieval is based on features (motion, spatial, visual, syntactic
or semantic characteristics of objects)
- Operating in compressed domain (MPEG, not progressive)
\item More information:
http://www.ctr.columbia.edu/~sfchang
CECS Multimedia Communications and Visualization Laboratory