Paper Title
Web Image Re-ranking using Visual Semantic Spaces

Web image search engine is a tool using that large number of images are searched. With present search engines for given query keyword multiple images are retrieved based on textual information and after selecting query image remaining images are re-ranked based on visual similarities of query images. These visual similarities of images do not well correlate with query image semantic meaning. This is a main problem in visual semantic web based search engine. In this paper we propose novel image re-ranking framework using that image visual and semantic meaning are automatically learned and then visual features of images are projected into their semantic meaning to get visual semantic signatures. Using online step search images are re-ranked by comparing their visual semantic signatures which obtained from query image semantic meaning. Our proposed framework significantly improve accuracy and efficiency of image re-ranking. Visual semantic signatures are comparatively smaller than original visual features of image on average. Index Terms - Image search ,Image re-ranking, Keyword expansion, and Visual semantic signature.