Visual semantic web based image Re-ranking for effective search engine

Abstract Visual semantic web based Image search engine is a way using that multiple images are search and matched in semantic space. This matched images we use for image reranking methodology. Image re-ranking is a method using that we improve results of web based images search. When user search any query keyword on web based search engine, then a set of images are extracted based on the textual information. User then select a required query image from the set of images and then the others images are recomputed or re-ranked based on visual occurrence of the query image. These similarities of visual features do not well match with visual semantic meanings of images which normally coordinate users search intention and it is a main problem visual semantic web image search engine. So In this paper, we developed visual semantic web image re-ranking structure, which automatically and directly offline studied different visual semantic spaces meaning for different search query keywords. Then these visual features of images are extended to their visual semantic spaces to formed visual semantic signatures. At the online step these images are ranked by comparison with their visual semantic signatures which obtained from the semantic space that specified by the search query keyword. Our proposed visual semantic web based image re-ranking structure significantly improve both the efficiency and accuracy of image re-ranking.