Ontology Based Image Retrieval in the Compressed Domain

Suhendro Irianto, Boby Bachry

Abstract: Image Ontology based image retrieval has been presented in this paper, the work also present visualization approach that aims to building effective image retrieval with high accuracy. In order to calculate the accuracy of image retrieval, the work used a method called accuracy classified. Accuracy classified can be defined as item classified correctly divided by multiply by 100 %. In this research DataGenie used ODBC to connect to a MySQL database. The research demonstrates that accuracy for each class greater than 50 % except for cat and emu classes which have of Life Object Super Class. In additional to demonstrating potential for high accuracy, the Image ontology proposed can also be used to visualize the relationship of each instance with its slot. The useful feature of protg demonstrated in this work was its ability to migrate any relational database into it. Furthermore the image ontology can also be saved as XML or HTML files so they can be used by any web application

Keywords: Image ontology, Protg, jambalaya, DataGenie