Combining visual and textual features for indexing images: a hybrid approach

Abstract

In this paper, we propose a new hybrid approach for indexing images and retrieving images from a corpus of images .The proposed approach is based on the fusion of the low level visual and textual descriptors. First, we describe each image of corpus in a model based on graphs. Then, in a second model, we translate the legends associated to images. These two models permit us to calculate the set of visual and textual descriptors for global index .We uses heuristics for the calculation of these features. The proposed approach is tested on a set of images and the results are compared with the human expert indexing.

Link http://www.ijitcs.com/volume%204_No_2/Ali+Benafiabanta.php