Hybrid Method to obtain Interest Region and Non Interest Region for Color Based Image Retrieval

Mamat, Abd. Rasid and Abdul Rawi, Norkhairani and Awang, Mohd Isa and Abdul Kadir, Mohd Fadzil and Abd Rahman, Mohd Nordin (2015) Hybrid Method to obtain Interest Region and Non Interest Region for Color Based Image Retrieval. International Journal of Advances in Soft Computing and its Application, 7 (3). pp. 1-15. ISSN 2074-8523 (In Press)

[img] Text
3IJASCA-070303_Pg16-30_Hybrid-Method-to-Obtain-Interest-Region-and-Non-Interest-Region.pdf
Restricted to Registered users only

Download (549Kb) | Request a copy
Official URL: http:\\www.home.ijasca.com/

Abstract

Content based image retrieval (CBIR) has become one of the most active research areas in the past few years. Many indexing techniques are based on global feature distributions. However, these global distributions have limited discriminating power because they are unable to capture local image information. In this paper, the new proposed method based on local image to classify the Interest Region (IR) and Non Interest Region (NIR) of images. To develop this, the integration of clustering and user intervention was applied. Clustering process is obtaining several regions, meanwhile to ascertain the location of the center of images through user intervention. Several experiments are conducted using different weight (ω, γ) of IR and NIR. Subsequently average color moment is extracted from this region (IR and NIR) in CIE Lab color model. To investigate the performance, new distance is proposed based on Euclidean distance. Experimental results show the proposed method more efficient in image retrieval. .

Item Type: Article
Keywords: Interest Region, Non Interest Region, Intervention, Clustering, CBIR
Subjects: Q Science > QA Mathematics > QA76 Computer software
Faculty / Institute: Faculty of Informatics & Computing
Depositing User: abd rasid mamat
Date Deposited: 21 Dec 2015 02:25
Last Modified: 21 Dec 2015 02:25
URI: http://erep.unisza.edu.my/id/eprint/3844

Actions (login required)

View Item View Item