Design And Realization Of Ontology-Based Image Retrieval System

Along with the appearance of Semantic Web, bring forward the perspective of the combination artificial intelligence network with human world. Some concept reference model in cultural and art field passes through many years' development, giving birth to the rock-bottom ontology that aims at cultural heritage and can apply on the domain knowledge of museum. At the same time,with the Beijing 2008 Olympic Games being near, the construction of digital museum in "Digital Olympic Games" plan performs very important. Traditional Chinese Painting is the gem of Chinese traditional arts. Automatically analyzing and understanding digitized art images and extracting their type, style and other semantic information is an important and imperative problem that needs to be addressed, it gets more and more attention.This dissertation emphasize that under the framework of Semantic Web, we use ontology technology to annotate cultural relic and retrieve them.Compared to the traditional retrieval methods,ontology-based retrieval can reduce "semantic gap" between human and computer harmonious interaction, and searching for a more comprehensive and accurate rate.My work is as follows:1) Analyzing the query requirement of the art images, and confirm the classified manner about the painting information. To combine with ontology standard, build semantic structure of domain knowledge and finish metadata design on the domain ontology model.2) Using seven layer architecture as the foundation for the Semantic Web , making ontological metadata to annotate art images and building a Knowledge Base. Then using Inference Engine to discover and analyse knowledge of the Knowledge Base.3) Working out a Bei-Hong Xu's digital museum architecture based on CIDOC CRM and ABC ontology. The system should classify paintings in Bei-Hong Xu's digital museum according to the inquiry demand of users, and storage them after analyzing them, which depend on the semantic image characteristics. The system offer the inquiry method which is on base of relation and characteristics .At last, using the framework of J2EE (Struts+Spring), the development tool of ontology editor and Jena the one of Semantic Web to prove the feasibility of this architecture.