A Novel Biologically Inspired Approach for Clustering and Multi-Level Image Thresholding: Modified Harris Hawks Optimizer

Biologically inspired computing deals with complex real-world problems using elegantly modeled techniques motivated by the behaviors of creatures in nature. Harris hawks optimizer (HHO), motivated by the cooperative behavior and hunting style of Harris' hawks, is a nature-inspired optimization paradigm. As an eminent swarm intelligence method, HHO has established strong performance. However, the original HHO may face difficulties when handling practical multimodal and composition problems. To overcome these challenges, this paper investigates an improved HHO, which considers nonlinear decay energy, introduces the grey wolf optimizer (GWO) as a competitive method to modify conventional HHO, and improves the balance between its exploration and exploitation. The proposed approach combines different cognitive hunting behaviors of Harris' hawks and grey wolf packs. The main idea of the proposed method can be described as follows: First, we generate a set of candidate solutions and then divide them into two halves. The improved HHO is employed to update the solutions in the first half, while the search phase of GWO is introduced to update the solutions in the second half. Second, we choose the best solutions for the union subpopulations and continue to conduct the iteration procedure. Furthermore, the new approach is utilized to solve the clustering problem and determine the optimal threshold values for multi-level image segmentation problems. Experimental results on 11 benchmark functions illustrate the effectiveness of the proposed approach. Extensive results on clustering and multi-level image segmentation demonstrate the efficiency of the proposed algorithm.

Automated summary

Biologically inspired computing is a method that uses elegantly modeled techniques motivated by the behaviors of creatures in nature to solve real-world problems. This paper investigates an improved Harris hawks optimizer (HHO) by introducing the grey wolf optimizer (GWO) and improving the balance between exploration and exploitation. The proposed approach combines different cognitive hunting behaviors of Harris' hawks and grey wolf packs and selects the best solutions through iterations. Experimental results demonstrate the effectiveness and efficiency of the proposed method.