<p>Hyperspectral Imaging (HSI), with its high spectral resolution, poses significant challenges in data processing due to the large volume of information. Recently, unsupervised metaheuristic-based band selection methods have gained popularity. However, these methods often neglect spatial information and suffer from inefficient optimization. To enhance the efficiency of band selection by joint exploitation of spatial and spectral information, this study proposes a novel Spatial Spectral Pufferfish Algorithm (SSPA) for unsupervised band selection in hyperspectral images. We employ a superpixel-based Fisher score as the objective function in the SSPA, effectively capturing both spatial homogeneity and spectral discriminability. The pufferfish optimization approach is used because it provides a robust mechanism to balance exploration and exploitation, helping to avoid local optima. The proposed SSPA was evaluated using three benchmark hyperspectral image datasets: Indian Pines, Pavia University, and Pavia Centre. Comparative analysis with five state-of-the-art band selection techniques demonstrates the superiority of our approach. The results indicate that SSPA improves the selection of informative bands and enhances the overall classification accuracy, validating its effectiveness and robustness. Our findings suggest that the integration of spatial context in spectral analysis significantly benefits unsupervised band selection, making SSPA a promising tool for hyperspectral image processing.</p>