Image retrieval with a multi-modality ontology.
Date of Issue2009
School of Computer Engineering
Centre for Multimedia and Network Technology
Ontology represents domain concepts and relations in a form of semantic network. Different from the traditional at structured feature vectors, ontology provides more information through concept de nition and relation inference. Many research works use ontologies in information matchmaking and retrieval. This trend is further accelerated by the convergence of various information sources supported by ontologies in the multimedia research FIeld. EFForts have been made to shift traditional content-based retrieval approaches toward concept-based approaches. Both text and image features are extracted and integrated to improve the classification and retrieval performance. However, there are many open issues on the ontology understanding, definition, construction, utilization and implementation. The extra work required by ontology based approaches becomes one of the major di culties that hedge against the development. This thesis mainly focuses on nding an e ective ontology model for multimedia information retrieval. More specifially, we look at the problem of image retrieval in the dynamic and noisy web environment, and propose a multi-modality ontology to better understand web image by incorporating both image and text features. Prototype systems are set up to prove the feasibility of the model. We then investigate the problem of scalability in ontology construction. Later, by taking advantage of both structural and content features of the online encyclopedia Wikipedia , real world objects are formalized in terms of concepts and relationships. Association rule mining algorithm is designed to improve the quality of the generated ontology. The retrieval performance by the automatically built ontology is comparable to the previous manually built one.
DRNTU::Engineering::Computer science and engineering::Information systems::Information storage and retrieval