Academic Journals Database
Disseminating quality controlled scientific knowledge

A New Segmentation Technique by KDE Model and Two-Phase Graph Cut Method

ADD TO MY LIST
 
Author(s): S. Bhagawath | T.D. Subha | A. Arul Mary

Journal: International Journal of Computer Science and Mobile Computing
ISSN 2320-088X

Volume: 2;
Issue: 4;
Start page: 628;
Date: 2013;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: kernel density estimation | Color saliency | graph cut | saliency model | salient object segmentation | seed adjustment | spatial saliency

ABSTRACT
In this paper, we propose a new segmentation approach by kernel density estimation (KDE) modeland two-phase graph cut method. Initially we have to construct set of KDE models by using the input imagepre-segmentation results. Then set of likelihoods for each pixels of all KDE models are calculated. Thesaliency map is generated by using color saliency and spatial saliency of each KDE models. Based on thecolor distinctiveness and spatial distributions the pixel-wise saliency map is generated. By using the twophase graph-cut method, the saliency map for complex image is generated. In the first phase, graph cut forinitial segmentation result is obtained. In the second phase, the final segmented image is generated.Experimental results for this technique produces quality image with good resolution and better performancefor 1000 test images.
Affiliate Program      Why do you need a reservation system?