Boomerang aerodynamic ellipse optimizer: A human game-inspired optimization technique for numerical optimization and multilevel thresholding image segmentation

The proliferation of image data has driven advances in segmentation methods, among which metaheuristic algorithms have emerged as a leading approach. In this paper, we introduce a novel metaheuristic optimizer inspired by the aerodynamic behavior of a boomerang in flight, explicitly modeling how release angle and launch force shape its trajectory. To overcome the limited local search capacity of existing algorithms, this paper proposes a uniform local mining strategy grounded in the aerodynamic ellipse effect and embed this mechanism within the boomerang motion model to create the boomerang aerodynamic ellipse optimizer. Evaluation on the CEC2017 benchmark functions reveals that the proposed optimizer consistently outperforms six recent comparison algorithms by achieving lower fitness values and faster convergence while maintaining robust performance across a range of problem dimensionalities. Application to image threshold segmentation on the BSDS500 dataset demonstrates superior FSIM, SSIM and PSNR metrics and stable segmentation quality across varying threshold counts, thus confirming the algorithm’s adaptability and reliability in practical image segmentation tasks. The boomerang aerodynamic ellipse optimizer therefore represents a significant contribution to metaheuristic-based image segmentation research. In summary, the boomerang aerodynamic ellipse optimizer emerges as a viable meta-heuristic optimization algorithm for image segmentation tasks. Source codes of the boomerang aerodynamic ellipse optimizer are publicly available at https://ww2.mathworks.cn/matlabcentral/fileexchange/181443-boomerang-aerodynamic-ellipse-optimizer.