The following information was submitted:
Transactions: WSEAS TRANSACTIONS ON SIGNAL PROCESSING
Transactions ID Number: 31-970
Full Name: Abdelilah Jilbab
Position: Doctor (Researcher)
Age: ON
Sex: Male
Address: GSCM_LRIT FSR University Med V Agdal-Rabat
Country: MOROCCO
Tel:
Tel prefix:
Fax:
E-mail address: a_jilbab@yahoo.fr
Other E-mails:
Title of the Paper: classification of natural and artificial images based on a maximum entropy
Authors as they appear in the Paper: Abdelilah Jilbab, Mohamed Daoudi, Dris Aboutajdine
Email addresses of all the authors: a_jilbab@yahoo.fr
Number of paper pages: 10
Abstract: A classification phase is crucial before the development of an image retrieval tool in the World Wide Web. To accelerate this classification and reduce the associated noise, we split first the images set in two semantic types, natural and artificial. This categorization is based on the selection of several features derived from the color and texture information that discriminates natural images from artificial ones. Usually, this features selection is performed in an empirical way and needs huge amount of experimentations. In order to avoid such experimentations and to represent efficiently the visual contents using a small number of features, we propose a new approach for evaluating their relevance. First, we select some features that could classify the images in two types: natural and artificial. Then, we elaborate a technique, based on the maximum entropy principle, to measure the relevance for each one of them. Hence we select the most relevant, and consequent!
ly improve the classification results.
Keywords: Image Classification; Maximum Entropy; Features Relevance
EXTENSION of the file: .doc
Special (Invited) Session:
Organizer of the Session:
How Did you learn about congress:
IP ADDRESS: 62.251.188.67
