Share Email Print
cover

Journal of Electronic Imaging

Entropy classification and discrete-cosine-transform-based image indexing system
Author(s): Yung-Gi Wu; Je-Hung Liu
Format Member Price Non-Member Price
PDF $20.00 $25.00

Paper Abstract

With the rapid growth of multimedia technology, more and more multimedia content is disseminated in networks or stored in databases. Image data is one of the multimedia types to be seen or accessed by users on the Internet or from databases. Searching the related images by querying image content is helpful for the management and usage of an image database. Therefore, research on image indexing techniques is an important topic. We propose an efficient content-based image retrieval (CBIR) system. Basically, the proposed method is a discrete cosine transform (DCT)-coefficient-based technique that extracts content features using some DCT coefficients. In addition, our method also uses entropy to classify images in the database so that it can reduce the search space to decrease the processing time. The proposed system has the property of robustness to rotation, translation, cropping, noise corruption, etc. The indexing time is only about 4 to 10% compared to most recently published results. According to our experiment results, the system is highly efficient in terms of robustness, precision, and a processing speed.

Paper Details

Date Published: 1 April 2006
PDF: 9 pages
J. Electron. Imaging. 15(2) 023019 doi: 10.1117/1.2194480
Published in: Journal of Electronic Imaging Volume 15, Issue 2
Show Author Affiliations
Yung-Gi Wu, Leader Univ. (Taiwan)
Je-Hung Liu, Leader Univ. (Taiwan)


© SPIE. Terms of Use
Back to Top