Unformatted text preview: of the data must be
reviewed again, adding new keywords as required. This could be a formidable task
even for small databases.
Content-based retrieval/querying (CBR/CBQ). In this method, queries are
based on the content oi multimedia data. When the data enters the database, it is
subjected to data analysis to derive certain attributes of the data (called derived
data). The derived data could include attributes such as lines, shapes, colors, and
textures from which objects could be determined at some later time. Queries
against this data then take place on the derived data. For example, "Find all
graphic objects having red triangles and green circles." Similarly, "Find all video
clips having a flying aeroplane."
To distinguish CBR on temporal data such as video and audio, from that on
images and graphics. CBR on images and graphics are called static CBR (or
SCBR), and CBR on video, animation, and audio are called dynamic CBR (or
DCBR). With both types of CBR, specifying precise queries (exact matching) generally is
not practical. Therefore, multimedia database query languages must have features
to support fuzzy predicates such as "Looks like this" to find approximate matches.
Furthermore, since queries might not yield exact matches, we need facilities for
ranking the retrieved results according to how closely they match the given query.
With the ranking facility, the multimedia database system should also support the
facility to browse the various retrieved data items, and the facility to retrieve
similar items based on one or more of the already retrieved items.
3. Query by example (QBE). QBE in case of multimedia database systems is a
special user interface to support CBR. These interfaces allow the user to specify
the attributes of the desired object(s) by providing examples to match such as, a
drawing, photograph, action, or sound. For example, the user can" draw an
example of the type of image to be retrieved, by using shape, color, or texture.
The query manager then us...
View Full Document
This document was uploaded on 04/07/2014.
- Spring '14