28.02.2013 Views

Bio-medical Ontologies Maintenance and Change Management

Bio-medical Ontologies Maintenance and Change Management

Bio-medical Ontologies Maintenance and Change Management

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

76 L. Stanescu, D. Dan Burdescu, <strong>and</strong> M. Brezovan<br />

1.4 Content-Based Image Retrieval Systems<br />

There is in present a series of database management systems capable to manage<br />

different types of media. Many of these systems permits indexing <strong>and</strong> searching<br />

for multimedia information taking into account only structured information, using<br />

traditional techniques. They work well only with short numeric <strong>and</strong> alphanumeric<br />

arrays. This includes traditional queries of the databases based on alphanumerical<br />

arrays. They cannot be used for multimedia information indexing <strong>and</strong> retrieval.<br />

That is why the researchers studied the possibility to create new systems that<br />

satisfy the high dem<strong>and</strong>s of the multimedia information.<br />

An exception is Oracle Multimedia (Formerly Oracle inter-Media), a feature of<br />

Oracle Database that enables the efficient management <strong>and</strong> retrieval of image,<br />

audio, <strong>and</strong> video data. Oracle Multimedia has knowledge of the most popular<br />

multimedia formats <strong>and</strong> makes automate metadata extraction <strong>and</strong> basic image<br />

processing [13, 50, 73].<br />

Most of the <strong>medical</strong> informatics systems use for multimedia <strong>medical</strong> databases<br />

management traditional database management servers (MySQL, MS SQL Server,<br />

Interbase, <strong>and</strong> Oracle). There have been implemented alternative methods for<br />

content-based visual retrieval, taking into consideration different characteristics<br />

like color, texture <strong>and</strong> shape.<br />

For <strong>medical</strong> multimedia databases content-based query it is generally used the<br />

method called QBE (Query by example). It implies the selection of an image or a<br />

region as a query image (region). For improving the results, the simple text-based<br />

query <strong>and</strong> content-based visual query can be combined [67].<br />

In general, the images are represented in the databases by automatically<br />

extracted visual features that are supposed to correspond to the visual image<br />

content or the way we perceive it. The features mainly used for image retrieval,<br />

are [30, 67]:<br />

• Grey levels <strong>and</strong> color descriptors, in a local or global fashion<br />

• Texture descriptors<br />

• Shapes of segmented objects<br />

Content-based retrieval has been investigated in a number of important<br />

projects. It can be mentioned the QBIC project from IBM [1, 26, 103] <strong>and</strong> Virage,<br />

a commercial project for content-based retrieval [103]. Most of the projects are<br />

academically projects: Photobook system from MIT [103], MARS (Multimedia<br />

Analysis <strong>and</strong> Retrieval System) developed to the University of Illinois [79, 103],<br />

Chabot system for image retrieval [101], WebSeek system, VisualSEEk <strong>and</strong> SaFe<br />

implemented to the University of Columbia [83, 84, 87, 103]. Using higher-level<br />

information, such as segmented parts of the image for queries, was introduced by<br />

the Blobworld system [9].<br />

A system that is available free of charge is the GNU Image Finding Tool<br />

(GIFT). Some systems are available as demonstration versions on the web such as<br />

Viper, WIPE or Compass [67].

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!