| < Day Day Up > |
|
Since the first multimedia database system was developed in 1987 (ORION), [260] the area and applications have experienced tremendous growth. As described in the previous section, all popular commercial database systems propose multimedia extenders to store, access, and query multimedia material and its metadata. Especially with the rapid development of network technology, multimedia database systems get more tremendous development, and multimedia information exchange becomes very important. Here, the reliance on standard technology is crucial to guarantee the interoperability of a multimedia database system and, thus, its commercial success. Hence, from the multimedia database viewpoint, the following issues have to be considered.
A multimedia data model must deal with the issue of representing the multimedia objects; that is, designing the high and low-level abstractions of the media data to facilitate various operations. These operations may include media object selection, insertion, editing, indexing, browsing, querying, retrieval, and communication. A good data model should fulfill the following issues:
It is necessary to model and store media components in the database. Its storage mechanism will play an important factor affecting the performance of a multimedia system.
A representation should be provided for the logical media structure. It is essential to represent this structure explicitly both for querying and for representation.
Semantic meaning should be modeled and linked to low-level characteristics and the structure of the media.
The metadata necessary for the operation of the system components needs to be determined and stored in the database.
It is necessary to rely on international standards, for example, MPEG-7, to guarantee interoperability for data sharing and data exchange.
The key functionality in a multimedia database is how to access and how to exchange multimedia information effectively. No matter what data model is used or data store mechanisms are employed, the most important thing is how to retrieve and communicate continuous and noncontinuous media with a short real-time constraint. The key to the retrieval process is similarity between two objects. The content of the object is analyzed and used to evaluate specified selection predicates, a process called contentbased retrieval. For instance, features described for visual retrieval include measures expressing the color distribution of the image or video. Other features express the texture and the composition of the image. An image query is translated into a point query in some multidimensional feature space. The similarity between a query and a database object is estimated using a distance function. The efficiency of the similarity search can be significantly improved by the use of multidimensional indexing structures.
To retrieve multimedia data from a database system, a query language must be provided. A multimedia query language must have the ability to handle complex spatial and temporal relationships. A powerful query language should have to deal with keywords, an index of keywords, and the semantic contents of multimedia objects.
The key to efficient communication is to rely on standards for communicating metadata and associated media data. An important contribution is the development of the system parts of MPEG-7 and the file format of MPEG-21.
Finally, the issues considered above on data model, storage retrieval and query should not be limited to atomic multimedia data such as image, audio, or video. It is also interesting to retrieve composite objects, such as when we read news text and, at the same time, open a window to see video documentaries. Hence, we need the ability to deal with several atomic multimedia data and media data presentations at the same time. MPEG-21 goes in this direction with the provision of the notion of a Digital Item, which surely will influence the multimedia database world in the future. In addition, and related again to MPEG-21, it is important for a multimedia database system to use multiple representations of data for different users and profiles, for intellectual property management and for adaptation purposes. In this context, the next chapter will detail technologies used in a distributed multimedia database environment.
[260]Woelk, D., Kim, W., and Luther, W., An object-oriented approach to multimedia databases, in SIGMOD Conf., Washington, D.C., 1986, pp. 311–325.
| < Day Day Up > |
|