Abstract—In this paper, the authors propose a novel automatic image annotation approach which relies on two main components: (i) identification of homogenous image regions, which share the same semantics using fuzzy clustering algorithm, and (ii) membership-based cross media relevance model to learn the association between keywords and image regions. The proposed fuzzy version of the Cross Media Relevance Model (CMRM) yields promising results. They use standard image collection to compare their approach to the original CMRM. The obtained results show that the proposed approach outperforms the original technique.

Index Terms—image annotation, unsupervised learning, fuzzy logic, image retrieval

Cite: Mohamed Alkaoud, Ibrahim AshShohail, and Mohamed Maher Ben Ismail, "Automatic Image Annotation Using Fuzzy Cross-Media Relevance Models," Journal of Image and Graphics, Vol. 2, No. 1, pp. 59-63, June 2014. doi: 10.12720/joig.2.1.59-63