改进水母优化算法及其在多阈值图像分割中的应用

Improved Jellyfish Search Optimizer Algorithm and Its Application in Multi-Threshold Image Segmentation

摘要: 针对水母优化算法(Jellyfish Search Optimizer, JSO)在高维优化问题中收敛速度慢、易陷入局部最优的缺陷，提出一种改进水母优化算法(Improved Jellyfish Search, IJS)。在JSO基础上引入自适应时间控制策略以平衡全局探索与局部开发能力，设计精英引导机制提升收敛精度与寻优效率，并采用硬边界约束保证迭代解在可行域内的有效性。为验证IJS算法的性能，本文首先基于6个标准测试函数，将IJS与JSO、灰狼优化算法、粒子群优化算法、鲸鱼优化算法进行寻优性能对比；其次将IJS应用于图像多阈值分割，基于Otsu类间方差最大化准则构建适应度函数，将FSIM、SSIM、PSNR作为评价指标，与JSO及传统Otsu算法对比分割效果。实验结果表明，IJS算法具有更优的寻优精度、收敛速度和稳定性，可有效应用于函数寻优与图像多阈值分割等复杂优化场景。

Abstract: To address the shortcomings of the Jellyfish Search Optimizer (JSO), including slow convergence and a tendency to become trapped in local optima when solving high-dimensional optimization problems, an Improved Jellyfish Search (IJS) algorithm is proposed. Based on the original JSO, an adaptive time control strategy is introduced to balance global exploration and local exploitation, an elite-guidance mechanism is designed to improve convergence accuracy and optimization efficiency, and a hard boundary constraint is adopted to ensure the validity of iterative solutions within the feasible domain. To verify the performance of the proposed IJS algorithm, six benchmark functions are first employed to compare its optimization performance with that of JSO, Grey Wolf Optimizer, Particle Swarm Optimization, and Whale Optimization Algorithm. Furthermore, IJS is applied to multi-threshold image segmentation, where the fitness function is constructed based on Otsu’s maximum between-class variance criterion. FSIM, SSIM, and PSNR are selected as evaluation metrics, and the segmentation performance is compared with that of JSO and the conventional Otsu method. Experimental results show that the IJS algorithm achieves superior optimization accuracy, convergence speed, and stability, and can be effectively applied to complex optimization scenarios such as function optimization and multilevel image thresholding.