Web-based Image Search using Query-Specific Semantic Signatures

Image re-ranking is a useful method for web-based image search. The search based on only s queried by the users is not efficient and results in imprecise output. The web-based image search used by Bing and Google uses image re-ranking. In image re-ranking, users' intention is captured by one-click on the query image. This helps in providing better search results to the users. In this paper, we review the method in which a query is first used to retrieve a plethora of images based on the. Image re-ranking framework automatically learns different semantic spaces offline for different query s. To get semantic signatures for images, their visual features are projected into their related semantic spaces. Images are re-ranked by comparing their semantic signatures and the query during the online stage. The query-specific semantic signatures, in the reviewed paper, significantly improve both the accuracy and efficiency of the re-ranking process. Hence, it is proved to be a better method than the conventional web-based image search techniques.