Kraisak Kesorn

Contact Details

Title: Research Student
Tel: Internal: [13] 7408
National: 020 7882 7408
International: +44 20 7882 7408
Fax: National: 020 7882 7997
International: +44 20 7882 7997
Email: kraisak [dot] kesorm [at] elec.qmul.ac.uk

Room: Eng 351

Research Group: Networks Research Group

Supervisor: Dr. Stefan Poslad

Project title: Cross-Linked Multi-Semantic Multi-Mode Annotations for Images Retrieval

Biography

Kraisak is a student from Thailand and start his PhD at Electronic Engineering Department, Queen Mary University of London in Oct, 2006. Image retrieval system is his research area under supervision of Dr.Stefan Poslad.

Project description

The quantity of visual information is expanding dramatically, resulting in many huge visual information databases. The visual media become a widespread information format in the World Wide Web (WWW). The sheer volume and variety of visual data can make searching more complex and time-consuming [Zhang, 2007]. Along with the quickly increasing demands to create and store visual information causes the need for searching facilities. Providing tools for effective access, retrieval, and management of huge visual information data, especially image and video, has attracted significant research efforts.

An inexperienced user may find that it is difficult to find the information she wants; even an experienced user may miss relevant Web pages. Users searching the WWW have a number of options presently available to them. Lycos, Excite, Alta Vista, and Yahoo! are named but few examples of useful search engines. All of these systems have been designed primarily to find text-based information on the Web. However, since images on the Internet are often poorly labeled, the keyword-based technique for image searching frequently yields poor results [Barnard, 2003]. Yet, over the past decade, ambitious attempts have been made to make computers learn to understand, index and annotate pictures representing a wide range of concepts, their relationships and context. One vital problem with all current approaches is the reliance on low-level visual similarity for judging semantic similarity, which may be problematic due to the semantic gap between low-level content and higher-level concepts.

This project aims to create the image retrieval system which is able to search images by the concepts rather than keywords may lead to better and more accurate visual data recall by users.