Combining image similarity metrics for semantic image annotation Host Publication: On the Move to Meaningful Internet Systems: OTM 2012 Workshops Authors: B. Jansen, D. Toan Tran and F. Temmermans Publisher: springer link Publication Date: Sep. 2012 Number of Pages: 3 ISBN: 978-3-642-33617-1
Abstract: This paper describes automated image annotation as an image retrieval problem, in which the distance metric used to express similarity among images is learnt from available distance metrics on several image descriptors (MPEGlj and SURF). Rather than describing the problem as an optimization problem, we study it as a regression problem. On a dataset of images of buildings taken in the city center of Brussels, we illustrate the superior performance of the combined dis- tance metrics over any of the considered individual distance metrics in automated image annotation.
|