Title: MR brain image segmentation using elite kinetic-molecular theory optimisation algorithm
Authors: Chaodong Fan; Ningjun Zheng; Juan Zou; Leyi Xiao
Addresses: Hunan Provincial Key Laboratory of Intelligent Information Processing and Application, Hengyang, Hunan, China; College of Information Engineering, Xiangtan University, Xiangtan, Hunan, China ' College of Information Engineering, Xiangtan University, Xiangtan, Hunan, China ' College of Information Engineering, Xiangtan University, Xiangtan, Hunan, China ' College of Electrical and Information Engineering, Hunan University, Changsha, Hunan, China
Abstract: This paper explores the use of the elite kinetic-molecular theory optimisation algorithm (EKMTOA) to compute threshold selection for MR brain image segmentation. EKMTOA is inspired by learning and collaboration among the elites individual. To solve the limitation of Otsu method based on projection of cross section that single threshold segmentation. Firstly, this paper proposes a multi-threshold cross section projection Otsu method. Secondly, adopt the post-processing strategy to further improve anti-noise capability. Finally, the EKMTOA is used to optimise the objective function, find the best segmentation threshold of the MR brain image. Experimental results demonstrate over multiple brain image with different of noise validate the efficiency of the proposed technique with regard to segmentation accuracy, anti-noise capability and robustness.
Keywords: image segmentation; cross section projection; EKMTOA; Otsu method.
DOI: 10.1504/IJAAC.2020.110074
International Journal of Automation and Control, 2020 Vol.14 No.5/6, pp.593 - 611
Received: 20 Oct 2018
Accepted: 06 Jun 2019
Published online: 05 Oct 2020 *