Review: Automatic image annotation for semantic image retrieval

Abstract

Nowadays, the number of digital data sets grows exponentially. Hence, the need to conceive efficient and powerful image indexation and retrieval systems grows as well. Automatic image annotation was adopted by several research as the emerging trend in image retrieval area. Actually, it is considered as the best solution that combines the content-based techniques by using low-level image features and text-based techniques exploiting textual annotations, associated to the image. In this way, the semantic gap between low-level image features and high-level semantics will be reduced. This paper presents a review of image retrieval approaches, by focusing especially on the automatic image annotation methods, in order to analyse the impact of annotations and associating semantics to the visual data for an image retrieval process.