Investigation of Multimedia Indexing Scheme on the Web

Kithangodage Jayaratne


ABSTRACT

Content-based indexing of multimedia has always been a challenging task. 
The enormity and diversity 
of the multimedia content on the World Wide Web (WWW) adds another dimension 
to this challenge. 
Today multimedia elements are increasingly being 
embedded in Web documents, and are being actively 
used to enhance the description of the document content. 
Since such documents over the WWW provide 
a rich source of information, the use of multimedia elements 
in Web documents has become very prevalent. 

Digital images have become increasingly indispensable in 
the world we live today, with the Internet 
offering a very rich variety of such both interesting and 
valuable images. However, managing such 
images to facilitate their retrieval is still under intensive 
research and is far from being perfect. With the 
increase in Internet bandwidth and CPU processing speed, 
the use of images has vociferated extensively 
in Web pages. Images are now frequently used to enhance the 
description of the content, capture the 
attention of readers and reduce the textual content of Web pages.  
In short digital images have become a 
very prominent component of the Web society today. 
To search for relevant images from such a large 
collection of images over the Web calls for a decisive 
mechanism that would exploit fully the semantics 
of the images. Unfortunately, most image search engines on the 
Web fail to exploit these image 
semantics, resulting poor recall and precision performance. 
It is thus one of our purposes here to 
propose an image representation model on a unified framework 
that would perform the relevance 
feedback on both the images' semantic content and the low level 
visual features such as color, shape and 
texture.