SEMANTIC WEB QUERY IMAGE RE - RANKING:ONLINE AND OFFLINE FRAMEWORK

Web image search engine is image re - ranking to improve the search results it is very significant feature. The paper can overcome the problem and new technique can introduced for web - s image re - ranking, online and offline with help of visual feature and mainly semantic signature of image. The mentioned technique is very useful in giving spe cific results to users in just one click. Image Search engines mostly use keywords and they depend on surrounding text for searching images. Duplication of image query is hard to describe accurately by using keywords. E g: Mouse is query keyword then categ ories can be "computer mouse", "white mouse","animals" etc. Given a query keyword, the various images can be present based on textual information. By asking the user to select a query image from the collection of images, the remaining images are re - ranked based on their visual similarities with the query image In different semantic spaces for different query keywords can be found offline database itself and automatically. Semantic signatures of the images are acquired by formatted by their visual features i nto their related semantic spaces and these semantic signatures. The visual and textual features of images are then considered into their related semantic spaces to get semantic signatures. In online stage images are re - ranked by comparing semantic signatu res obtained from semantic space obtained from query keywords.