Latest Issues on Multimedia Database
The Canadian Institute for
Telecommunication Research (CITR) Broadband Services studies and prototypes
enabling technologies for distributed multimedia applications using
Object-Oriented Multimedia database system. Such applications include
multimedia news, distance learning, and home shopping. A successful product
of their project is the News-on-Demand Application. This application
incorporates a distributed database server by storing up to date news items
from various sources, such as TV, radio, newspapers, and magazines.
News-on-Demand allows subscriber to retrieve one or more news items using an
article query interface. Various news items from different sources are
annotated and organized into multimedia documents by service providers.
News-on-Demand uses a disturbed service where a client can access news over
the broadband network.
Multimedia technologies
are attracting more and more interest every day. Video-on-Demand is one of
the buzzwords today and is now available for the public. Content providers
such as publishers, broadcasting companies and audio/video production firms
must be able to archive and index their productions for later retrieval.
This is a formidable task and it is even more so when the material to be
handled encompasses several media types and covers a time span of several
years. In order for such a vast amount of data to be easily available,
existing multimedia database design models, indexing and retrieval
methodologies and delivery methods have to be improved and refined. In
addition, video, image and audio data cannot be effectively managed with the
exclusive use of older, keyword-based search techniques.
The UCLA Knowledge-Based
Multimedia Medical Distributed Database (KMeD) project is a joint project
between the Computer Science Department and the Radiological Science
Department. The KMeD project has five major objectives:
-
Query medical multimedia distributed
database by image, alphabetical, and numerical content.
-
Model temporal, spatial, and evolutionary
nature of medical objects.
-
Formulate queries using conceptual and
imprecise medical terms to support cooperative processing.
-
Develop a domain-independent, high-level
query language and a medical domain user interface to support KMeD
functionality.
-
Provide analysis and presentation methods
for visualization of knowledge and data models.
KMeD database is presented by
features and objects. The selected objects of interest in medical images
such as X-Ray and MRI image are segmented using knowledge-based model-guiding
techniques. As with other multimedia database system features and contents
of the medical image are extracted and stored in a feature and content
database. In KMeD, Type Abstraction Hierarchies (TAHs) is used to represent
the knowledge about the image in three structures. A query process supports
operators such as "similar to" and "nearby" and conceptual terms such as
"small" and "large" to find the approximate matches to the features and
contents. A technique has been developed for visual interface to use
point-click-and-drag input. This new development by UCLA, namely KMeD
database, can change the way hospitals maintain and utilize all of their
data.
EtherMed
is an experimental database to Internet accessible multimedia courseware
in health professions education. EtherMed contains records of educational
materials that are freely accessible on the web and links to the actual
courseware. EtherMed records are descriptive and non-evaluative. EtherMed is an experiment is
collaborative database development and maintenance. OHPCC staff and
collaborating consultants at distant locations can directly enter, modify,
and delete records online. The aim is determine whether distributed
management of database resources is possible and can become
self-sustaining and to determine how well database approaches compare to
those employing search engines. Online videoconferencing and other
collaboration tools are used to support this effort.
ImageMed
is an experimental system that uses:
(1) the Unified Medical Language System (UMLS) Meta-thesaurus for indexing
and retrieval of medical images.
(2) a distributed image database.
Future Work
Future requirements of
distributed multimedia systems will be even much more demanding than it is
now. It is envisaged that users will be heavily mobile and require
ubiquitous access to, and satisfactory presentation of, multimedia data,
regardless of the actual connectivity and the specific presentation device
they are currently using (e.g., wire-based vs. wireless network,
high-resolution graphics screen vs. Web-enabled cellular phone). Moreover,
users will expect to find information and multimedia contents faster and
easily, and will interact with the contents much more intensively than
today.
In this scope, the CODAC
and its sister project aims at realizing a quality adaptive end-to-end
multimedia system, i.e., we shall provide means for indexing and retrieving
multimedia data by their content and adaptation capabilities and develop
methods to guarantee a quality adaptive video transport to the client.
Depicts the
architectural view of the end-to-end multimedia system, which we would like
to realize. For this we need to carry out a number of activities and the
following are the first work items under development in the CODAC project:
1-Development of a multimedia cartridge in the core of an
Oracle 9i DBMS: The advantages of the cartridge technology, as proposed by
Oracle and other database vendors, are reusability, extensibility and
especially a very clean interface to components of the database system, like
query processor, optimizer and page access management. This multimedia cartridge
realizes the
meta-database and provides access to the clients for complex search
functionality, supported by advanced
indexing structures (like a combination of X-trees, SS-trees, etc.).
2.
Realization of a
Processing Unit for
MPEG-4 videos: The Processing Unit is supposed to be situated
between the
video server and the meta-database and shall extract the necessary quality
adaptation capability information from the A/V streams to be stored as
meta-data in the database. Upon the insertion of the video (the insertion
could be on demand or on a regular basis), the Processing Unit shall apply
efects such as transformation and scaling to the MPEG-4 encoded videos and
report results (performance and variation information) to the meta-database
and write back the adapted videos to the video server.
Implementation of an
indexing structure for the access of MPEG-7 files, possibly in BiM format:
A composition mechanism of different MPEG-7 Access Units shall be developed
and the mapping process of MPEG-7 Access Units to MPEG-4 Access Units shall
be implemented.
4. Realization of the
cross-referencing between MPEG-7 and MPEG-4, i.e., how to access media data
from meta-data and vice-versa: indexing structures of the later work item
shall be employed and integrated into a reference stream which allows the
efficient access of the meta-data from the media-data. Referencing of MPEG-7
and MPEG-4 will be employed in the proxy-cache and in the active routers.
The application
scenarios, we focus upon are sport event videos, M3box, and tele-teaching.
Sport events are an interesting application scenario since these videos
provide semantically rich content. The M3box is an adaptive multimedia
message box. It is developed by Siemens Corporate Technology.
2) In cooperation with
our Institute. The final application is tele-teaching. This application
differs from the previous ones, as here, multimedia data plays a supporting
role, rather than being the central data component. Therefore, the
meta-database has to keep, besides descriptive information on the multimedia
data, information on the tele-teaching material and the teaching process.
Future
Improvement
The most interesting
and exciting thing about multimedia databases is how quickly they're
evolving. This growth, along with the emergence of inexpensive removable
storage devices such as DVD-which stores tens of gigabytes-will ignite an
explosion of multimedia applications. This explosion, in turn, will fuel an
intense need for powerful multimedia databases.
|