Welcome!
To use the personalized features of this site, please log in or register.
If you have forgotten your username or password, we can help.
My Menu
Saved Items

Regular Paper

Scalable representation of vector descriptors

Dong-Gyu SimContact Information and Hae-Kwang Kim2

(1) artment of Computer Engineering, Image Processing Systems Laboratory, Kwangwoon University, 447-1, Wolgye-dong, Nowon-gu, Seoul, 139-701, South Korea
(2) Department of Software Engineering, Sejong University, Yulkok-kwan Room No. 403C, 98 Kunja-dong, Kwangjin-gu, Seoul, South Korea

Published online: 9 March 2006

Abstract  In this paper, a general bit-plane representation method is presented to achieve effective scalable representation for vector descriptors. Conventional vector descriptors are represented by a sequence of scalar elements and the scalability can be achieved by changing the number of elements. The proposed method represents the vector descriptors by a sequence of bit-planes that are obtained by grouping the values at the same bit positions of all the elements. In this method, the scalability is achieved by changing the number of the bit-planes. We applied the proposed scalability feature to the MPEG-7 texture descriptor and show that the performance of the proposed representation is significantly better than that of conventional representations in terms of retrieval accuracy and stability with a coarse representation.

Keywords  Indexing - Scalability - Multimedia - Descriptor - MPEG-7 - Retrieval


Contact InformationDong-Gyu Sim
Email: dgsim@kw.ac.kr
Fulltext Preview (Small, Large)
Image of the first page of the fulltext

References secured to subscribers.



Export this article
Export this article as RIS | Text
 
Remote Address: 38.107.191.113 • Server: mpweb22
HTTP User Agent: CCBot/1.0 (+http://www.commoncrawl.org/bot.html)