Optimized tag ranking based on visual vocabulary for social images in compressed domain

With the rapid development of social media, research about its application and organization has attracted considerable attention. Many social media sharing websites allow users to tag the uploaded images with tags. However, the tags are usually uncontrolled, ambiguous, and overly personalized. One of the most challenging issues is how to match the textual tags with the visual image to rank tags accurately. Because a great deal of social images are stored and transmitted in a compressed format, an optimized tag ranking based on visual vocabulary in compressed domain is proposed in order to improve the accuracy of tag ranking and further reduce the ranking time. First, the low-resolution social images are constructed from the compressed image data. Then a visual vocabulary is created after the SIFT descriptor is extracted. Finally, the neighbor voting model is utilized to rank the image tags. Experimental results show that the proposed optimization method can significantly reduce the time of image tag ranking under ensuring the ranking accuracy of social image tags.