1 / 9

Multimedia Databases

Multimedia Databases. How to represent and organize the databases containing a mix of media types? MM DB architectures The Principle of Autonomy: Use suitable index structures for each media type (may result in different index structures)

joey
Download Presentation

Multimedia Databases

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Multimedia Databases • How to represent and organize the databases containing a mix of media types? • MM DB architectures • The Principle of Autonomy: Use suitable index structures for each media type (may result in different index structures) • The Principle of Uniformity: Use same index structure for all media types • The Principle of Hybrid Organization: Hybrid of the above two • Organizing based on the principle of uniformity • associate with each media object oi, some meta-data, md(oi) • md(oi) may be produced by a human or program (image/video/text content retrieval engine) • If our archive contains objects o1, … on, then index the metadata md(o1), … md(on) in a way that provides efficient ways of implementing the expected accesses that users will make

  2. Media Abstractions • Considering the content of mm data of different media types, What is common (or different) to all the media types • Media abstractions are mathematical structures representing such media content • Media abstractions may be implemented through a single data structure

  3. Media Abstractions • Media abstraction comprises of • a set of states (state is the smallest chuck of media that we wish to reason about) • a set of features (feature is any object in a state that is deemed to be of interest, in principle can include both objects and activities) • attributes associated with features • a feature extraction map that specifies what features occur in which states • in some cases, these are implemented as content extraction programs • in some other cases may involve humans manually specifying content • a set of state-dependent and state-independent relations • e.g., left-of is state-dependent, age is state-independent • a set of inter-state relations • e.g. before(s1,s2) may say that s1 occurred before s2

  4. Media Abstractions: Image data • States: {pic1.gif, … pic7.gif} • Features: Names of the people shown in the photographs (Bob, Jim, Bill, Charlie, Ed) • Extraction Map: • state feature • pic1.gif Bob,Jim • pic2.gif Jim • pic3.gif Bob … • Relations • state-dependent : left-of • state-independent: father • Inter-state relations may be empty

  5. Media Abstractions: Video data • States: frames 1 .. 5 • Features: Jane Shady, Denis Dopeman, Dopeman-house, briefcase, .. • Extraction Map: • state feature • frame1 Jane Shady, Dopeman-house, briefcase • frame2 Jane Shady, Denis Dopeman, Dopeman-house, briefcase • frame3 Jane Shady, Denis Dopeman, Dopeman-house, briefcase • frame4 Jane Shady, Denis Dopeman, Dopeman-house, briefcase • frame5 Jane Shady, Dopeman-house

  6. Media Abstractions: Video data • Relations • state-dependent : have • person object state • Jane Shady briefcase 1 • Jane Shady briefcase 2 • Jane Shady briefcase 3 • D.Dopeman briefcase 4 • state-independent: spouse • person spouse • Jane Shady Peter Shady • Peter Shady Jane Shady • Inter-state relations: • before(s1,s2) holds iff s1 occurs before s2

  7. Simple Multimedia Database • Is a finite set of media extractions • simple multimedia databases are naive • a media abstraction may list “church” as a feature • when searching for “cathedrals” or “monuments” we may not find church because the system does not know that they are synonymous • users often search for media objects containing one or more features and then “refine” the search later when the returned objects (though correct) do not correspond to what they wanted

  8. Structured Multimedia Database • SMDS consists of • set of media abstractions M • equivalence relation ( )on features F • partial ordering () of the set on F/equivalence classes on F • inh: a map that associates with each feature f, a set of features below f according to the ordering on features • subst: is a map from

  9. Example of SMDS • media object part/frame features • image photo1.gif - church,danube • image photo2.gif - cathedral,melk • image photo3.gif - church,st.paul, rome • video video1.mpg 1-5 church,stream • video video1.mpg 6-10 stream • audio audio1.wav 1-20 st.peters,tiber • 3 media abstractions, one each associated with image, video and audio • features F = church, danube, cathedral, melk, st.paul, rome, stream, st.peters, tiber • equivalence • church  cathedral • river  stream • partial ordering  church cathedral river stream St.paul St.peter danube Tiber rome melk

More Related