An Efficient Image Retrieval Model Using Fuzzy Semantic Concepts

Abstract

Concepts can add knowledge to the interpretation of image contents. However, mapping low-level features to high-level image semantics is still an ongoing challenge for researchers. In this paper an integrated model of fuzzy centrality and intensity concepts, together with the concept hierarchy is proposed to efficiently retrieving images. The self-organization feature map is applied to construct a three-layer concept hierarchy for image archives. Thus, search for the image concepts can be effectively achieved by detecting the presences of the relevant bottom-level image primitive features. In other words, an image can be categorized into multiple semantics. Consequently, the retrieval accuracy can be improved by searching the multiple categories. The methodology of the proposed model will be illustrated in this paper and the experimental results will be presented to demonstrate the efficiency in retrieving images.