Probabilistic image tagging with tags expanded by text-based search

Abstract

Automatic image tagging automatically assigns image with semantic keywords called tags, which significantly facilitates image search and organization. Most of present image tagging approaches assign the query image with the tags derived from the visually similar images in the training dataset only. However, their scalabilities and performances are constrained by the limitation of using the training method and the fixed size tag vocabulary. In this paper, we proposed a search based probabilistic image tagging algorithm (CTSTag), in which the initially assigned tags are mined from the content-based search result and expanded from the text-based search results. Experiments on NUS-WIDE dataset show not only the performance of the proposed algorithm but also the advantage of image retrieval using the tagging result. © 2011 Springer-Verlag.