Title: Global minimisation of fuzzy level set for image segmentation

Authors: Guoqi Liu; Chenjing Li; Ming Deng

Addresses: College of Computer and Information Engineering, Henan Normal University, Xinxiang, China ' College of Computer and Information Engineering, Henan Normal University, Xinxiang, China ' College of Computer and Information Engineering, Henan Normal University, Xinxiang, China

Abstract: Level set is an important method in image segmentation, and some models based on level set method have obtained great success, such as Chan and Vese (C-V) and its convex formulation, local binary fitting (LBF) model. However, these models have two drawbacks to be simultaneously solved. One is the non-convexity of energy functional; the other difficulty is segmenting objects in the background of inhomogeneous intensity. In order to simultaneously cope with these shortcomings, a fuzzy level set energy functional model is proposed. Firstly, a fuzzy factor is introduced in the original LBF model to describe the intensity inhomogeneity. Besides, the edge information is also integrated into the proposed model to improve the robustness of extracting objects. Finally, a regularisation optimisation method is introduced to obtain the global minimisation. Experimental results confirm the proposed method is robust to initialisation and could segment objects with inhomogeneous intensity.

Keywords: image segmentation; fuzzy level set; global minimisation; intensity inhomogeneity.

DOI: 10.1504/IJWMC.2018.092357

International Journal of Wireless and Mobile Computing, 2018 Vol.14 No.3, pp.209 - 215

Received: 09 Jun 2017
Accepted: 06 Dec 2017
Published online: 16 Jun 2018 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article