Image re-ranking using sift feature
Nita A. Jaybhaye, Sudhir G. Shikalpure · 2016
To improve the web based result there is effective way is to use image re-ranking. Which has been used in many professional search engines like Google and Bing. Given a query keyword the images are retrieved based on that query keyword. By asking the user to select one image from pool of retrieved images other images are re-ranked based on their visual similarities. A major challenge is that according to users search intention similarity of visual feature are not well correlate in semantic meaning. In this paper, we propose image re-ranking framework using SIFT features, which groups similar images into clusters. Each cluster contains similar images based on visual and textual features. At the online stage, images are retrieved and re-ranked. The proposed query-specific semantic signatures mostly take least time with respect to other and get better accuracy image search and re-ranking.