A hybrid transient search naked mole-rat optimizer for image segmentation using multilevel thresholding. (1st March 2023)

• Record Type: Journal Article
• Authors: Singh, Supreet; Mittal, Nitin; Nayyar, Anand; Singh, Urvinder; Singh, Simrandeep
• Is Part Of: Expert systems with applications. Volume 213:Part B(2023)
• Publication Date: 2023-03-01
• Subjects: Numerical optimization -- Naked mole rat algorithm -- Transient search optimizer -- Adaptive properties -- Segmentation -- Image thresholding; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2022.119021
• Languages: English
• ISSNs: 0957-4174

Abstract

Abstract: This paper introduces a new hybrid algorithm named Transient Search Naked Mole-Rat Optimizer (TsNMRO), based on the hybridization of transient search optimizer (TSO) merged in the naked mole-rat algorithm (NMRA). To improve the worker phase of traditional NMRA, the new algorithm combines TSO's exploratory notions with NMRA. A novel stagnation phase motivated by the grey wolf optimizer (GWO) and cuckoo search (CS) algorithms has been implemented to tackle the local optima stagnation problem. In addition, self-adaptivity has been included by adjusting all of the proposed algorithm's parameters. For varied parameters of the algorithm under examination, a simulated annealing ( s a ) based mutation technique has been used, requiring no user-based parametric modifications. CEC 2005, CEC 2019 benchmark challenges, and image thresholding issues are identified to evaluate the TsNMRO's performance. The significance of the suggested technique across higher dimensional problems and variable function evaluations has been demonstrated by a comparison of varying dimension sizes. It can be observed from the experimental outcomes that the suggested TsNMRO outperforms competitive methods in terms of mean and standard deviation values for most of the cases. The suggested TsNMRO's improved performance is further validated by statistical results such as Freidman's test, Wilcoxon's rank-sum test and convergence profiles. Highlights: Hybridization of TSO and NMRA to develop the TsNMRO. A stagnation phase based on GWO-CS algorithms is employed for local optimal solution. The proposed TsNMRO is made to be self-adaptive. Performance evaluation for CEC2005 and CEC2019 numerical test problems. Real world optimization problem of Image Segmentation.