Image Segmentation Using Mrf Novel Level Set Method

K. Srilatha, S. Pavithra, Neenu Vincent, S. Venkatesh, A. S.Vigneshwar and Medapati Jahnavi Lakshmi

Department of Electronics and Communication engineering Sathyabama University, Tamilnadu, Chennai, India.

Corresponding Author E-mail: ¹srilatha169@gmail.com

DOI : http://dx.doi.org/10.13005/bbra/2463

ABSTRACT: The most difficult task in the segmentation of the image is to obtain the segmentation by avoiding almost all the noises within less span of time. The images which are extracted will be from any domain i.e., medical, natural, radar etc. In the proposed method the MRF (Markov random field) algorithm for the process of segmentation is implemented. In this algorithm two methods in parallel fashion are used. The two methods which are used is AMG (Algebraic multi grid) and SFM (Sparse random field).The AMG is used for increasing the time step and SFM will be used in decreasing the computation domain. In the novel level set method it considers that the neighboring pixels also fall in the same region. The noises also will be reduced in this method when it is compared with the existing methods. By using this method, the images of 500*500 sizes also can be segmented. In this technique the number of iterations will be less when compared to the existing systems. The images of many domains can be segmented by using this proposed algorithm i.e., Medical images, noisy images, synthetic aperture radar images (SAR) and natural images. The SAR, natural and noisy images are applicable only in the real-time.

KEYWORDS: Algebraic multigrid (AMG); contour evolution. Novel level set method Markov random field (MRF); sparse random field (SFM);

[ HTML Full Text]