Intelligent Semantic Web Image Search Engine

Search engine services are popular means of information retrieval. Search engines perform a number of tasks to retrieve required information from web based on their respective architecture. This paper proposes a novel approach for web crawling and image retrieval based on the image re-ranking strategy by considering both textual and visual features of the image. A pool of images will be retrieved when the user gives a query, then user has to select a query image which is similar to the target image from the initial search results. Users search intention can be identified using this one click user feedback. The query image is then fed to a SVM classifier that identifies the reference classes by analyzing the features of the image. Then the images are re-ranked using rankSVM algorithm by considering the query and click data. This image searching strategy has improved accuracy and efficiency.