Multilevel thresholding is recognized as a fast and effective technique for image segmentation. Although exhaustive search provides a comprehensive solution, its computational complexity increases with the number of threshold levels. This paper introduces a novel meta-heuristic search algorithm called Ensemble Searching (ES), designed to tackle complex nonlinear optimization problems. The focus is on applying ES to image multilevel thresholding. Initially, the population is divided into predefined groups, each guided by an evolutionary algorithm that independently searches for better positions within the search space. If an algorithm encounters a local optimum, a diversity-maintaining mechanism is activated to relocate the group. Throughout the iterative process, all algorithms share the best global solution (Gbest). The proposed structure’s effectiveness is evaluated using ten test images and the energy curve method. Kapur’s entropy, a well-established measure, is used to assess the algorithm’s performance. A comparative analysis with eight different search algorithms demonstrates the proposed framework’s rapid convergence, confirming its efficiency and effectiveness.
