Optimization of Remote Sensing Image Segmentation Based on Improved Harris Hawks Algorithm

ABSTRACT Remote sensing imagery requires accurate segmentation in many applications. However, multilevel thresholding is computationally expensive. As the number of thresholds increases, the search space expands rapidly, and segmentation performance becomes sensitive to complex backgrounds and noise. This study proposes an improved Harris Hawks optimization (HHO)‐based approach for multithreshold remote sensing image segmentation. Three strategies are integrated to strengthen the exploration–exploitation balance. These modifications accelerate convergence and improve segmentation quality. Experiments on remote sensing images from the NWPU‐RESISC45 dataset show that the proposed method outperforms representative swarm intelligence baselines, including JSOA, SMA, MPA, BOA, WOA, and SSA in statistical indicators. Under noise, it maintains stable performance, demonstrating improved robustness. Overall, the proposed optimizer provides an efficient and robust solution for multilevel thresholding in remote sensing imagery.