Narrowing the semantic-gap using multi-modal ontology for semantic image retrieval

摘要

An image can be represented by its low level visual feature or by its context. The context kind of image representation requires the intelligence of human annotation which would not be unique. The main objective of this paper is to address the problem of semantic gap and procedure to narrow it. The gap between the human context description and visual feature description of the image is said to be semantic gap. In this paper the concept of ontologies are used to narrow the semantic gap. Here the low level features such as prevalent color of image, basic intrinsic pattern, and contour gradient are used to address the image. This concept is implemented for an Asteroideae flower family domain.