Tag-based Image Retrieval using Hybrid Visual-Tag Feature Extraction Method

Abstract

The rapid development of digital cameras and use of image sharing websites generates a number of images and retrieving an image from such large database is an important task.  In an image sharing social website tags around the images provides valuable information for searching and classification of the images. In this paper the Hybrid Visual-Tag Feature Extraction method (HVT-FE) is proposed for retrieval of tagged images. The HVT-FE includes color moment, wavelet packet transform and Tag Frequency-Inverse Image Frequency (TF-IIF) features. The HVT-FE gives better result performance as compared to the existing feature extraction techniques.

Keywords: Tag, Image Features, Flickr, Tag-Feature based Image Retrieval.