Tools
Tools
Zerzour, Kamel (2005) A semantic model for content and event-based indexing and retrieval of video databases. Doctoral thesis, London Metropolitan University.
The representation, indexing and retrieval of semantic video content is an intensively investigated research area that requires efforts from several fields of computer science, notably, Information Retrieval, Database and Artificial Intelligence communities. A principled approach to the description of a video model requires a formal representation of video data, addressing two dimensions: form (static) and their semantic (dynamic). Correspondingly, there are three categories of video indexing; one for each dimension and one concerning the combination of both. This thesis proposes a semantic object-oriented video data model for representing and indexing video at both the form- and semantics-based features. The model makes full and proper use of the semantics and knowledge in dealing with the representation and indexing of video data. More importantly, all forms of indexing, and hence retrieval, coexist in a well-founded framework, which combines in a neat way different techniques to represent video data, exploiting further the notion that video modelling is a multi-disciplinary task. To model the form-based features, Object-Oriented (OO) database modelling techniques are employed, while Description Logic (DLs) models the semantics-based features. To increase the usability of the model, the integration of semantics- and form-based features is addressed in the shape of a schema manager that allows interpretation of form-based features onto semantic features, and facilitates the migration of semantic feature onto the permanent storage provided by the underlying OODMS database management system. A prototypical implementation, VIGILANT, model has been developed in a full OO environment using the Objectivity/DB database management system. Both form- and semantics-based features are implemented using OO techniques. The DLs knowledge base is simulated as an OO database exploiting the similarities across DLs and OO databases.
 |
View Item |