Interpretation of user intention via textual and visual resemblance

The speedy evolution in web environment and progression in technology have led us to access and manage tremendous images easily in various areas. Current internet image search engines purely trust on the text based information around the images. Keywords supplied by user can not specify content of images exactly. Returned images composed of many noisy, doubtful and insignificant images. To solve above confusion in text-based image retrieval, it is beneficial to utilize visual details of image. System has been proposed to overcome the uncertainty of images such that users intention can be determined by one click internet image search. User selects a query image among image pool retrieved by extended text-based search. Our approach contributes excess clusters created by using candidate words and visual content of images by Scale Invariant Feature Transform (SIFT) algorithm. Weight of image is estimated by using strategy of adaption weight. Results are enriched by re-ranking of images by similarity measure calculation. Duplicate images are detected and removed by applying Message Digest (MD5) hash function on images. Quality of results is further enhanced by considering resolution of images. User intention is determined by integration of extended textual and visual resemblance without additional user energy. Experimental investigation demonstrates significant improvement in terms of user satisfaction and relevancy.