Contextual image search with keyword and image input

In this paper, we propose a novel image search scheme is contextual image search with keyword input. It is different from conventional image search schemes. it consist of three step process, first one is context extraction to distinguish the image entities of the same name, second step is conceptualization to convert context into a list of weighted concepts using wikification. Wikification is process, which links the noun phrases in plain text content to the corresponding articles third step is tri stage image clustering to process the conceptualized contexts. Those three stages are Tag context clustering, Text context clustering and Expansion clustering. Where Tag context clustering clusters with most reliable signal using HAC (Hierarchical Agglomerative Clustering), Text context clustering uses the text content of all images to form the cluster using HAC algorithm and Expansion clustering using inverse document frequency. Finally the web images are clustered. The contextual image search based on keyword system presents a separate interface (e.g., text input box) to allow users to submit a query; this system enables users to find the images based on different name entity images. Though the keyword based contextual image search provides result it doesn´t find the duplicate images. Thus the conceptual image search with image input concept is proposed to find the duplicate images. Contextual image search with input image, the users can select an clustered image as the search query from Web pages or other documents they are reading. Using MPEG-7 descriptors to “annotate” an image automatically. Finally it determine the duplicate images on search result.