Abstract:

Semantic extraction has been a challenging problem in multimedia databases. Widely used semantic extraction methods are based on manual annotation and/or low-level features (color, shape and texture). Low-level features usually don’t provide good results when considering high level semantics and manual annotation presents promising results but it is time consuming and will be impossible when considering the huge database compose of thousands of images. To cater the problem of multimedia semantics we have presented the concept of semantic libraries. The main focus of this paper is to explain how semantics libraries will interact to extract highlevel semantics. Semantics are extracted mainly by considering the dominant objects in the image and the relationships among these objects. Design and working along with example and case study of these semantic libraries have been shown