A Frequency–Spatial Domain Enhancement Method for Mountain Remote Sensing Images Based on Improved Multi-Objective Particle Swarm Optimization

Mountainous remote sensing images are often affected by terrain undulations and variations in solar incidence angles, resulting in uneven illumination, insufficient contrast, and blurred terrain textures. To address these challenges, this study proposes a frequency–spatial domain hybrid enhancement method based on an improved multi-objective particle swarm optimization (MOPSO). The method integrates homomorphic filtering with contrast-limited adaptive histogram equalization (HF–CLAHE) to construct a hybrid framework that balances global illumination correction with local texture enhancement. Key enhancement parameters are adaptively optimized using the improved MOPSO, achieving coordinated optimization of brightness uniformity, detail enhancement, and structural preservation. Typical mountainous Landsat 8 images are used for experiments and compared with several conventional enhancement methods and the traditional MOPSO-optimized HF–CLAHE. The results show that, compared with the traditional MOPSO-optimized HF–CLAHE, the proposed method achieves average improvements of 19.42%, 36.06%, 42.25%, and 24.96% in standard deviation, mean gray value, visual information fidelity, and mean gradient, respectively, effectively enhancing terrain texture details while preserving the original spatial structures.