European researchers working on the Cognitive-Level Annotation Using Latent Statistical Structure (CLASS) project are developing visualization technologies capable of recognizing both specific objects and classes of objects. "The recognition of an object as belonging to a particular group is a harder problem for a computer than the recognition of a specific object," says Luc Van Gool of Belgium's Leuven University.
"The reason is that object classes show large variability among their members." The CLASS project has developed a system in which the description of an object is based on the appearance of numerous, small patches. The researchers say that localized features provide the necessary robustness to handle the massive variations that occur within a group of objects. CLASS also developed a special mechanism, called efficient approximate neighborhood search, for the comparison of an image or an object with a vast number of reference images.
The CLASS project's technology has been incorporated into a commercial application that enables mobile phone subscribers to take pictures of specific objects and receive relevant information on the object's subject. "It's like the object itself becomes the link to further information," Van Gool says. He says cities and museums could use this technology to offer interactive guided tours.
From ICT Results
View Full Article
Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA
No entries found