13.4 Current Activities Involving Video Database Systems

Multimedia Information Systems with Video Images

(1) Computer-Aided Instruction System :( By Parkes)

Advantages: This model consists of PART-OF relationships among description data and binary relationships between settings.

Weakness: Did not address the issue of inheritance of description data.

 example:

(2) Muse in Project Athena:( By Hodges, Sasnett, and Ackerman)

Advantages: Muse is a useful system to compose multimedia applications.

Weakness: neither data model issues nor the associated mechanism fro handling description data of video information was mentioned.

(3) EVA :( By MIT)

Advantage: EVA is a useful tool with which to analyze video data.

Weakness: its capability to share description information among annotated video scenes is relatively weak.

(4) Harmony:

Harmony is a multimedia presentation system, and adopts the hyper_object model. ( including object-oriented model and hypertext model.)

Duo to the model of timed Petri net, synchronization among different media can be described.

Data Models for Video Databases

(1) Time-Interval-Based Model:( By Little and Ghafoor)

Compose multimedia data into 2 categories:

Spatial composition: the notion of spatial arrangement of data.

Temporal composition: how to maintain the synchronization of data in a process of data presentation.

(2) Temporal Logic for Time Intervals:( By Allen)

if a condition P holds during a time interval T, then P holds during any subinterval t of T.

(3) Virtual Museum:( By Gibbs)

Provided a data model which resembles the E-R data model. Video and sound data are considered as media objects in this model.

(4) Video Document Model (VDM) : ( By Hara)

This model is basically a hypertext model. The arbitrary scenes and objects in each scene are  defined as nodes. Each node can have its own attributes and values.

Video Retrieval Systems

(1) Video Indexing : ( By Tanaka)

A full video search system which automatically detects the flipping point of cuts in a given video stream and searches for a similar object in each scene.

(2) Automatic Video Database Building : ( By Ariki)

Build video database by using television news video.

Have employed the following automatic video processing techniques:

[a] Detection of scene flipping.

[b] Extraction of camera work such as panning or zooming.

[c] Keyword extraction from narration.

[d]  Keyword extraction from superimposed test.

Storage and Delivery Systems

(1) Digital TV Studio : ( By Maier)

Two notable points with regard to this system are:

[a] Constrained-Latency Storage Access ( CLSA) is a method which the system can control the delivery and display of video images in multimedia object ( video) retrieval.

[b] Describe the quality of the data. In each script, the quality of services ( QOS) for object is described.

(2) Video Server :( By Ghandeharizadeh)

 Propose several techniques for video servers. In particular, the notion of Object declustering is introduced as a way to reply video continuously, and to improve the response for multi-user access.