Social Sensing Image Rank using Visual Words Module I

Abstract

The proposed tag-based ranking algorithm for image retrieval uses both the meta-data of images and visual features of images to retrieve the most relevant result set. Hence, making the retrieval of images more accurate than the tag-based and contentbased image retrievals. Initially, the system extracts meta-data of images and stores them into a specially designed dictionary dataset. Then, given a textual query, the system computesdifferent relevance scores on the basis of user tags and content tags (provided to us from Module- II). Next, we combine the scores from both user tags and content tags to compute a final list of scores. At the end, we rank our images, which are related to the user query and display the result based on the new list of scores.